<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>DEV Community: Huyen Hai Nghi</title>
    <description>The latest articles on DEV Community by Huyen Hai Nghi (@huyentoolnotes).</description>
    <link>https://dev.to/huyentoolnotes</link>
    <image>
      <url>https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4251531%2Fdaf1e201-0953-433e-abac-90226e7a2f35.png</url>
      <title>DEV Community: Huyen Hai Nghi</title>
      <link>https://dev.to/huyentoolnotes</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/huyentoolnotes"/>
    <language>en</language>
    <item>
      <title>Validate OFX 1.02 SGML FITID in Python: A 5-Gate Script</title>
      <dc:creator>Huyen Hai Nghi</dc:creator>
      <pubDate>Sat, 21 Nov 2026 15:53:29 +0000</pubDate>
      <link>https://dev.to/huyentoolnotes/validate-ofx-102-sgml-fitid-in-python-a-5-gate-script-3b0</link>
      <guid>https://dev.to/huyentoolnotes/validate-ofx-102-sgml-fitid-in-python-a-5-gate-script-3b0</guid>
      <description>&lt;p&gt;To validate OFX 1.02 SGML FITID in Python without crashing on legacy bank exports, treat the file as a plain-text header plus an SGML token stream instead of XML. Track container tags on a stack, allow unclosed leaf tags, and run &lt;code&gt;FITID&lt;/code&gt; deduplication alongside fixed-point &lt;code&gt;Decimal&lt;/code&gt; balance checks after parsing.&lt;/p&gt;

&lt;p&gt;Most Python tutorials stop right after &lt;code&gt;pip install ofxparse&lt;/code&gt; or &lt;code&gt;ofxtools&lt;/code&gt; and print &lt;code&gt;tx.fitid&lt;/code&gt;. Both libraries deserialize bank rows fine, yet neither rejects a file whose &lt;code&gt;Opening + Deposits - Withdrawals&lt;/code&gt; total misses &lt;code&gt;&amp;lt;LEDGERBAL&amp;gt;&amp;lt;BALAMT&amp;gt;&lt;/code&gt; by &lt;code&gt;$142.50&lt;/code&gt;, nor stops two transactions inside the same statement from reusing an identical &lt;code&gt;&amp;lt;FITID&amp;gt;&lt;/code&gt;. Worse, an invalid or stripped &lt;code&gt;&amp;lt;INTU.BID&amp;gt;&lt;/code&gt; tag passes silently until accounting software rejects the Web Connect upload.&lt;/p&gt;

&lt;p&gt;Any preflight check for these exports operates under five constraints:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Python 3.10+ stdlib modules only (&lt;code&gt;re&lt;/code&gt;, &lt;code&gt;decimal&lt;/code&gt;, &lt;code&gt;datetime&lt;/code&gt;, &lt;code&gt;dataclasses&lt;/code&gt;, &lt;code&gt;pathlib&lt;/code&gt;)&lt;/li&gt;
&lt;li&gt;OFX &lt;code&gt;1.0&lt;/code&gt; through &lt;code&gt;1.6&lt;/code&gt; (&lt;code&gt;VERSION:102&lt;/code&gt; SGML headers), not XML-based OFX &lt;code&gt;2.x&lt;/code&gt; payloads&lt;/li&gt;
&lt;li&gt;Single-byte bank encodings (&lt;code&gt;CHARSET:1252&lt;/code&gt; as &lt;code&gt;cp1252&lt;/code&gt;, &lt;code&gt;ISO-8859-1&lt;/code&gt; as &lt;code&gt;latin_1&lt;/code&gt;, or &lt;code&gt;UTF-8&lt;/code&gt;)&lt;/li&gt;
&lt;li&gt;Unclosed leaf tags (&lt;code&gt;&amp;lt;TRNTYPE&amp;gt;&lt;/code&gt;, &lt;code&gt;&amp;lt;DTPOSTED&amp;gt;&lt;/code&gt;, &lt;code&gt;&amp;lt;TRNAMT&amp;gt;&lt;/code&gt;, &lt;code&gt;&amp;lt;FITID&amp;gt;&lt;/code&gt;, &lt;code&gt;&amp;lt;INTU.BID&amp;gt;&lt;/code&gt;)&lt;/li&gt;
&lt;li&gt;Explicit closing tags required on containers (&lt;code&gt;&amp;lt;OFX&amp;gt;&lt;/code&gt;, &lt;code&gt;&amp;lt;STMTRS&amp;gt;&lt;/code&gt;, &lt;code&gt;&amp;lt;BANKTRANLIST&amp;gt;&lt;/code&gt;, &lt;code&gt;&amp;lt;STMTTRN&amp;gt;&lt;/code&gt;)&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  Why standard XML parsers fail when you validate OFX 1.02 SGML FITID in Python
&lt;/h2&gt;

&lt;p&gt;Standard XML parsers fail as soon as you try to validate OFX 1.02 SGML FITID in Python because &lt;a href="https://en.wikipedia.org/wiki/Open_Financial_Exchange" rel="noopener noreferrer"&gt;OFX versions 1.0 through 1.6&lt;/a&gt; use SGML, not XML. Every valid file begins with an ASCII key-value block (&lt;code&gt;OFXHEADER:100&lt;/code&gt;, &lt;code&gt;DATA:OFXSGML&lt;/code&gt;, &lt;code&gt;VERSION:102&lt;/code&gt;) above &lt;code&gt;&amp;lt;OFX&amp;gt;&lt;/code&gt;, and data elements inside transactions omit closing tags.&lt;/p&gt;

&lt;p&gt;Pass that raw file to Python's &lt;a href="https://docs.python.org/3/library/xml.etree.elementtree.html" rel="noopener noreferrer"&gt;&lt;code&gt;xml.etree.ElementTree&lt;/code&gt;&lt;/a&gt;, and it aborts on line 7 with &lt;code&gt;ParseError: mismatched tag: line 7, column 2&lt;/code&gt;:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;OFXHEADER:100
DATA:OFXSGML
VERSION:102
SECURITY:NONE
ENCODING:USASCII
CHARSET:1252
COMPRESSION:NONE
OLDFILEUID:NONE
NEWFILEUID:NONE

&amp;lt;OFX&amp;gt;
  &amp;lt;SIGNONMSGSRSV1&amp;gt;
    &amp;lt;SONRS&amp;gt;
      &amp;lt;STATUS&amp;gt;
        &amp;lt;CODE&amp;gt;0
        &amp;lt;SEVERITY&amp;gt;INFO
      &amp;lt;/STATUS&amp;gt;
      &amp;lt;INTU.BID&amp;gt;3000
    &amp;lt;/SONRS&amp;gt;
  &amp;lt;/SIGNONMSGSRSV1&amp;gt;
  &amp;lt;BANKMSGSRSV1&amp;gt;
    &amp;lt;STMTTRNRS&amp;gt;
      &amp;lt;STMTRS&amp;gt;
        &amp;lt;CURDEF&amp;gt;USD
        &amp;lt;BANKTRANLIST&amp;gt;
          &amp;lt;DTSTART&amp;gt;20261001000000
          &amp;lt;DTEND&amp;gt;20261031235959
          &amp;lt;STMTTRN&amp;gt;
            &amp;lt;TRNTYPE&amp;gt;DEBIT
            &amp;lt;DTPOSTED&amp;gt;20261004141500
            &amp;lt;TRNAMT&amp;gt;-184.20
            &amp;lt;FITID&amp;gt;20261004002
            &amp;lt;NAME&amp;gt;PACIFIC ELECTRIC
          &amp;lt;/STMTTRN&amp;gt;
        &amp;lt;/BANKTRANLIST&amp;gt;
        &amp;lt;LEDGERBAL&amp;gt;
          &amp;lt;BALAMT&amp;gt;6465.15
          &amp;lt;DTASOF&amp;gt;20261031235959
        &amp;lt;/LEDGERBAL&amp;gt;
      &amp;lt;/STMTRS&amp;gt;
    &amp;lt;/STMTTRNRS&amp;gt;
  &amp;lt;/BANKMSGSRSV1&amp;gt;
&amp;lt;/OFX&amp;gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Open-source OFX parsers handle the SGML header and unclosed leaves, though reading their source code turns up two surprising validation gaps:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;a href="https://raw.githubusercontent.com/csingley/ofxtools/master/ofxtools/header.py" rel="noopener noreferrer"&gt;&lt;code&gt;ofxtools.header.OFXHeaderV1&lt;/code&gt;&lt;/a&gt; decodes headers and &lt;a href="https://raw.githubusercontent.com/csingley/ofxtools/master/ofxtools/Parser.py" rel="noopener noreferrer"&gt;&lt;code&gt;ofxtools.Parser.TreeBuilder&lt;/code&gt;&lt;/a&gt; closes leaf tags during tree construction, yet &lt;a href="https://raw.githubusercontent.com/csingley/ofxtools/master/ofxtools/models/base.py" rel="noopener noreferrer"&gt;&lt;code&gt;Aggregate.groom()&lt;/code&gt;&lt;/a&gt; drops every tag whose name contains a dot (&lt;code&gt;if "." in child.tag: elem.remove(child)&lt;/code&gt;). Both the &lt;a href="https://raw.githubusercontent.com/csingley/ofxtools/master/README.rst" rel="noopener noreferrer"&gt;ofxtools README&lt;/a&gt; and the &lt;a href="https://ofxtools.readthedocs.io/en/latest/parser.html" rel="noopener noreferrer"&gt;ofxtools parser guide&lt;/a&gt; confirm that Intuit proprietary extensions such as &lt;code&gt;&amp;lt;INTU.BID&amp;gt;&lt;/code&gt; get stripped with a warning rather than checked for valid bank IDs.&lt;/li&gt;
&lt;li&gt;
&lt;a href="https://raw.githubusercontent.com/jseutter/ofxparse/master/ofxparse/ofxparse.py" rel="noopener noreferrer"&gt;&lt;code&gt;ofxparse&lt;/code&gt;&lt;/a&gt; runs a regex preprocessor (&lt;code&gt;OfxPreprocessedFile&lt;/code&gt;) that collects every &lt;code&gt;&amp;lt;/TAG&amp;gt;&lt;/code&gt; name into &lt;code&gt;closing_tags&lt;/code&gt; first, then closes a preceding opening tag before the next tag only when that tag name is absent from &lt;code&gt;closing_tags&lt;/code&gt;. Because &lt;code&gt;STMTTRN&lt;/code&gt; appears in &lt;code&gt;closing_tags&lt;/code&gt; whenever other transactions close properly, omitting a single &lt;code&gt;&amp;lt;/STMTTRN&amp;gt;&lt;/code&gt; leaves that element open in the preprocessed stream rather than flagging a broken block on its own line.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Neither package enforces statement-wide &lt;code&gt;&amp;lt;FITID&amp;gt;&lt;/code&gt; uniqueness or verifies ledger totals against &lt;code&gt;&amp;lt;LEDGERBAL&amp;gt;&amp;lt;BALAMT&amp;gt;&lt;/code&gt;. That leaves you writing a thin validation pass before loading rows.&lt;/p&gt;

&lt;h2&gt;
  
  
  Building the five-gate OFX 1.02 validator step by step
&lt;/h2&gt;

&lt;p&gt;A preflight validator passes raw bytes through five ordered gates so structural breaks abort early while valid files reach balance math:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;
&lt;strong&gt;Decode headers and reject leading BOM bytes&lt;/strong&gt;: Split bytes at &lt;code&gt;&amp;lt;OFX&amp;gt;&lt;/code&gt;, flag any UTF-8 BOM (&lt;code&gt;\xef\xbb\xbf&lt;/code&gt;) ahead of &lt;code&gt;OFXHEADER:100&lt;/code&gt;, verify &lt;code&gt;DATA:OFXSGML&lt;/code&gt; and &lt;code&gt;VERSION:102&lt;/code&gt;, and decode strictly via &lt;code&gt;CHARSET&lt;/code&gt; (&lt;code&gt;cp1252&lt;/code&gt;, &lt;code&gt;latin_1&lt;/code&gt;, or &lt;code&gt;utf_8&lt;/code&gt;). Many US banks emit &lt;code&gt;ENCODING:USASCII&lt;/code&gt; with &lt;code&gt;CHARSET:1252&lt;/code&gt;, so strict &lt;code&gt;cp1252&lt;/code&gt; decoding catches broken bytes without dropping punctuation.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Walk SGML tokens with a container stack&lt;/strong&gt;: Match &lt;code&gt;&amp;lt;(/?)([A-Za-z0-9_.]+)&amp;gt;\s*([^&amp;lt;]*)&lt;/code&gt; sequentially, pushing container tags (&lt;code&gt;OFX&lt;/code&gt;, &lt;code&gt;STMTRS&lt;/code&gt;, &lt;code&gt;BANKTRANLIST&lt;/code&gt;, &lt;code&gt;STMTTRN&lt;/code&gt;, &lt;code&gt;LEDGERBAL&lt;/code&gt;) onto a stack while allowing unclosed leaves (&lt;code&gt;FITID&lt;/code&gt;, &lt;code&gt;TRNAMT&lt;/code&gt;, &lt;code&gt;INTU.BID&lt;/code&gt;). A second &lt;code&gt;&amp;lt;STMTTRN&amp;gt;&lt;/code&gt; opening before &lt;code&gt;&amp;lt;/STMTTRN&amp;gt;&lt;/code&gt; pops logs the open and failure line numbers cleanly.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Validate &lt;code&gt;&amp;lt;INTU.BID&amp;gt;&lt;/code&gt; and calendar dates&lt;/strong&gt;: Check that &lt;code&gt;&amp;lt;INTU.BID&amp;gt;&lt;/code&gt; (used by &lt;a href="https://en.wikipedia.org/wiki/Open_Financial_Exchange#Intuit_and_QFX" rel="noopener noreferrer"&gt;Intuit's QFX/QBO Web Connect extensions&lt;/a&gt;) is a positive integer, and validate &lt;code&gt;&amp;lt;DTPOSTED&amp;gt;&lt;/code&gt;, &lt;code&gt;&amp;lt;DTSTART&amp;gt;&lt;/code&gt;, and &lt;code&gt;&amp;lt;DTEND&amp;gt;&lt;/code&gt; against &lt;a href="https://raw.githubusercontent.com/csingley/ofxtools/master/ofxtools/Types.py" rel="noopener noreferrer"&gt;OFX Section 3.2.8.2 regex rules&lt;/a&gt; plus &lt;code&gt;datetime.date()&lt;/code&gt; so &lt;code&gt;20260230&lt;/code&gt; fails immediately.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Enforce &lt;code&gt;&amp;lt;FITID&amp;gt;&lt;/code&gt; uniqueness and &lt;code&gt;TRNTYPE&lt;/code&gt; polarity&lt;/strong&gt;: Require a non-whitespace &lt;code&gt;&amp;lt;FITID&amp;gt;&lt;/code&gt; within &lt;code&gt;255&lt;/code&gt; characters on every &lt;code&gt;&amp;lt;STMTTRN&amp;gt;&lt;/code&gt;, detect collisions locally or against a historical &lt;code&gt;set()&lt;/code&gt;, and flag sign reversals (&lt;code&gt;DEBIT&lt;/code&gt; with positive &lt;code&gt;&amp;lt;TRNAMT&amp;gt;&lt;/code&gt;, or &lt;code&gt;CREDIT&lt;/code&gt; with negative amounts).&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Reconcile balances in &lt;code&gt;decimal.Decimal&lt;/code&gt;&lt;/strong&gt;: Sum deposits and withdrawals using &lt;a href="https://docs.python.org/3/library/decimal.html" rel="noopener noreferrer"&gt;&lt;code&gt;decimal.Decimal&lt;/code&gt;&lt;/a&gt; instead of binary &lt;code&gt;float&lt;/code&gt;, then test &lt;code&gt;opening_balance + deposits - withdrawals == reported_closing&lt;/code&gt; against &lt;code&gt;&amp;lt;LEDGERBAL&amp;gt;&amp;lt;BALAMT&amp;gt;&lt;/code&gt;.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;Save the runnable implementation below as &lt;code&gt;validate_ofx.py&lt;/code&gt; and pass any &lt;code&gt;.ofx&lt;/code&gt; or &lt;code&gt;.qbo&lt;/code&gt; file on the command line:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight python"&gt;&lt;code&gt;&lt;span class="k"&gt;from&lt;/span&gt; &lt;span class="n"&gt;decimal&lt;/span&gt; &lt;span class="k"&gt;import&lt;/span&gt; &lt;span class="nc"&gt;Decimal&lt;/span&gt;
&lt;span class="k"&gt;import&lt;/span&gt; &lt;span class="n"&gt;datetime&lt;/span&gt; &lt;span class="k"&gt;as&lt;/span&gt; &lt;span class="n"&gt;dt&lt;/span&gt;
&lt;span class="k"&gt;from&lt;/span&gt; &lt;span class="n"&gt;pathlib&lt;/span&gt; &lt;span class="k"&gt;import&lt;/span&gt; &lt;span class="nc"&gt;Path&lt;/span&gt;
&lt;span class="k"&gt;import&lt;/span&gt; &lt;span class="n"&gt;re&lt;/span&gt;
&lt;span class="k"&gt;import&lt;/span&gt; &lt;span class="n"&gt;sys&lt;/span&gt;

&lt;span class="no"&gt;CONTAINER_TAGS&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="kt"&gt;frozenset&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;{&lt;/span&gt;
    &lt;span class="s2"&gt;"OFX"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"SIGNONMSGSRSV1"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"SONRS"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"STATUS"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"FI"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"BANKMSGSRSV1"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
    &lt;span class="s2"&gt;"STMTTRNRS"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"STMTRS"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"BANKACCTFROM"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"CREDITCARDMSGSRSV1"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
    &lt;span class="s2"&gt;"CCSTMTTRNRS"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"CCSTMTRS"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"CCACCTFROM"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"BANKTRANLIST"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"STMTTRN"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
    &lt;span class="s2"&gt;"PAYEE"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"LEDGERBAL"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"AVAILBAL"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"BALLIST"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"BAL"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"CURRENCY"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"ORIGCURRENCY"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
&lt;span class="p"&gt;}&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
&lt;span class="no"&gt;OUTFLOW_TYPES&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="kt"&gt;frozenset&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="s2"&gt;"DEBIT"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"FEE"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"SRVCHG"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"ATM"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"POS"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"CHECK"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"PAYMENT"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"CASH"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"DIRECTDEBIT"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"REPEATPMT"&lt;/span&gt;&lt;span class="p"&gt;}&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
&lt;span class="no"&gt;INFLOW_TYPES&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="kt"&gt;frozenset&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="s2"&gt;"CREDIT"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"DEP"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"INT"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"DIV"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"DIRECTDEP"&lt;/span&gt;&lt;span class="p"&gt;}&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
&lt;span class="no"&gt;CHARSET_MAP&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="s2"&gt;"1252"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s2"&gt;"cp1252"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"ISO-8859-1"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s2"&gt;"latin_1"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"8859-1"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s2"&gt;"latin_1"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"NONE"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s2"&gt;"utf_8"&lt;/span&gt;&lt;span class="p"&gt;}&lt;/span&gt;

&lt;span class="no"&gt;TAG_RE&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="n"&gt;re&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nb"&gt;compile&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;r&lt;/span&gt;&lt;span class="s2"&gt;"&amp;lt;(/?)([A-Za-z0-9_.]+)&amp;gt;\s*([^&amp;lt;]*)"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
&lt;span class="no"&gt;DT_RE&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="n"&gt;re&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nb"&gt;compile&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;
    &lt;span class="n"&gt;r&lt;/span&gt;&lt;span class="s2"&gt;"^(?P&amp;lt;year&amp;gt;\d{4})(?P&amp;lt;month&amp;gt;0[1-9]|1[0-2])(?P&amp;lt;day&amp;gt;0[1-9]|[12]\d|3[01])"&lt;/span&gt;
    &lt;span class="n"&gt;r&lt;/span&gt;&lt;span class="s2"&gt;"(?:(?P&amp;lt;hour&amp;gt;[01]\d|2[0-3])(?P&amp;lt;minute&amp;gt;[0-5]\d)"&lt;/span&gt;
    &lt;span class="n"&gt;r&lt;/span&gt;&lt;span class="s2"&gt;"(?:(?P&amp;lt;second&amp;gt;[0-5]\d|60)(?:\.(?P&amp;lt;ms&amp;gt;\d{3}))?)?"&lt;/span&gt;
    &lt;span class="n"&gt;r&lt;/span&gt;&lt;span class="s2"&gt;"(?:\[(?P&amp;lt;offset&amp;gt;[+-]?\d+(?:\.\d{2})?)(?::[^\]]+)?\])?)?$"&lt;/span&gt;
&lt;span class="p"&gt;)&lt;/span&gt;


&lt;span class="k"&gt;def&lt;/span&gt; &lt;span class="nf"&gt;parse_ofx_datetime&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;raw&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="kt"&gt;str&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="o"&gt;-&amp;gt;&lt;/span&gt; &lt;span class="n"&gt;dt&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="n"&gt;date&lt;/span&gt; &lt;span class="o"&gt;|&lt;/span&gt; &lt;span class="kc"&gt;None&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
    &lt;span class="n"&gt;m&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="no"&gt;DT_RE&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="k"&gt;match&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;raw&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;strip&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
    &lt;span class="k"&gt;if&lt;/span&gt; &lt;span class="ow"&gt;not&lt;/span&gt; &lt;span class="n"&gt;m&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
        &lt;span class="k"&gt;return&lt;/span&gt; &lt;span class="kc"&gt;None&lt;/span&gt;
    &lt;span class="k"&gt;try&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
        &lt;span class="k"&gt;return&lt;/span&gt; &lt;span class="n"&gt;dt&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;date&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="kt"&gt;int&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;m&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="s2"&gt;"year"&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="kt"&gt;int&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;m&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="s2"&gt;"month"&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="kt"&gt;int&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;m&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="s2"&gt;"day"&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
    &lt;span class="k"&gt;except&lt;/span&gt; &lt;span class="nc"&gt;ValueError&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
        &lt;span class="k"&gt;return&lt;/span&gt; &lt;span class="kc"&gt;None&lt;/span&gt;


&lt;span class="k"&gt;def&lt;/span&gt; &lt;span class="nf"&gt;decode_header&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;raw&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="kt"&gt;bytes&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="o"&gt;-&amp;gt;&lt;/span&gt; &lt;span class="kt"&gt;tuple&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="kt"&gt;str&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="kt"&gt;int&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="kt"&gt;list&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="kt"&gt;str&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
    &lt;span class="n"&gt;errs&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="kt"&gt;list&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="kt"&gt;str&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;
    &lt;span class="k"&gt;if&lt;/span&gt; &lt;span class="n"&gt;raw&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;startswith&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;b&lt;/span&gt;&lt;span class="s2"&gt;"\xef\xbb\xbf"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
        &lt;span class="n"&gt;errs&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;append&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="s2"&gt;"G1 line 1: UTF-8 BOM precedes OFXHEADER:100"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
        &lt;span class="n"&gt;raw&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="n"&gt;raw&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="mi"&gt;3&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;
    &lt;span class="n"&gt;idx&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="n"&gt;raw&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;lower&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;find&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;b&lt;/span&gt;&lt;span class="s2"&gt;"&amp;lt;ofx"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
    &lt;span class="k"&gt;if&lt;/span&gt; &lt;span class="n"&gt;idx&lt;/span&gt; &lt;span class="o"&gt;==&lt;/span&gt; &lt;span class="o"&gt;-&lt;/span&gt;&lt;span class="mi"&gt;1&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
        &lt;span class="k"&gt;return&lt;/span&gt; &lt;span class="n"&gt;raw&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;decode&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="s2"&gt;"utf_8"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;errors&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="s2"&gt;"replace"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="mi"&gt;1&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;errs&lt;/span&gt; &lt;span class="o"&gt;+&lt;/span&gt; &lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="s2"&gt;"G1 line 1: missing &amp;lt;OFX&amp;gt; root"&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;
    &lt;span class="n"&gt;hdr_txt&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="n"&gt;raw&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="n"&gt;idx&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;decode&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="s2"&gt;"ascii"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;errors&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="s2"&gt;"replace"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
    &lt;span class="n"&gt;hdr&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="n"&gt;k&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;upper&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="n"&gt;v&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;strip&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="k"&gt;for&lt;/span&gt; &lt;span class="n"&gt;k&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;v&lt;/span&gt; &lt;span class="k"&gt;in&lt;/span&gt; &lt;span class="n"&gt;re&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;findall&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;r&lt;/span&gt;&lt;span class="s2"&gt;"([A-Z]+)\s*:\s*([^\s\r\n]+)"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;hdr_txt&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;}&lt;/span&gt;
    &lt;span class="k"&gt;if&lt;/span&gt; &lt;span class="n"&gt;hdr&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;get&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="s2"&gt;"OFXHEADER"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="o"&gt;!=&lt;/span&gt; &lt;span class="s2"&gt;"100"&lt;/span&gt; &lt;span class="ow"&gt;or&lt;/span&gt; &lt;span class="n"&gt;hdr&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;get&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="s2"&gt;"DATA"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="o"&gt;!=&lt;/span&gt; &lt;span class="s2"&gt;"OFXSGML"&lt;/span&gt; &lt;span class="ow"&gt;or&lt;/span&gt; &lt;span class="n"&gt;hdr&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;get&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="s2"&gt;"VERSION"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="o"&gt;!=&lt;/span&gt; &lt;span class="s2"&gt;"102"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
        &lt;span class="n"&gt;errs&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;append&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;f&lt;/span&gt;&lt;span class="s2"&gt;"G1 line 1: invalid OFX 1.02 header {hdr}"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
    &lt;span class="n"&gt;enc&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="n"&gt;hdr&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;get&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="s2"&gt;"ENCODING"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"USASCII"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;upper&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
    &lt;span class="n"&gt;codec&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="s2"&gt;"utf_8"&lt;/span&gt; &lt;span class="k"&gt;if&lt;/span&gt; &lt;span class="n"&gt;enc&lt;/span&gt; &lt;span class="k"&gt;in&lt;/span&gt; &lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="s2"&gt;"UTF-8"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"UNICODE"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="k"&gt;else&lt;/span&gt; &lt;span class="no"&gt;CHARSET_MAP&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;get&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;hdr&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;get&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="s2"&gt;"CHARSET"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"NONE"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;upper&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"cp1252"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
    &lt;span class="k"&gt;return&lt;/span&gt; &lt;span class="n"&gt;raw&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="n"&gt;idx&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;decode&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;codec&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;errors&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="s2"&gt;"strict"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;hdr_txt&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;count&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="s2"&gt;"\n"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="o"&gt;+&lt;/span&gt; &lt;span class="mi"&gt;1&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;errs&lt;/span&gt;


&lt;span class="k"&gt;def&lt;/span&gt; &lt;span class="nf"&gt;validate_ofx_file&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;path&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="nc"&gt;Path&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;opening&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="nc"&gt;Decimal&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;known_fitids&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="kt"&gt;set&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="kt"&gt;str&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt; &lt;span class="o"&gt;|&lt;/span&gt; &lt;span class="kc"&gt;None&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="kc"&gt;None&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="o"&gt;-&amp;gt;&lt;/span&gt; &lt;span class="kt"&gt;list&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="kt"&gt;str&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
    &lt;span class="n"&gt;body_text&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;cur_line&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;errors&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="nf"&gt;decode_header&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;path&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;read_bytes&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
    &lt;span class="n"&gt;stack&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="kt"&gt;list&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="kt"&gt;tuple&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="kt"&gt;str&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="kt"&gt;int&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;
    &lt;span class="n"&gt;last_leaf&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;last_end&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="kc"&gt;None&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="mi"&gt;0&lt;/span&gt;
    &lt;span class="n"&gt;txs&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;current_tx&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;meta&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="kc"&gt;None&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="p"&gt;}&lt;/span&gt;

    &lt;span class="k"&gt;for&lt;/span&gt; &lt;span class="n"&gt;m&lt;/span&gt; &lt;span class="k"&gt;in&lt;/span&gt; &lt;span class="no"&gt;TAG_RE&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;finditer&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;body_text&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
        &lt;span class="n"&gt;gap&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="n"&gt;body_text&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="n"&gt;last_end&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="n"&gt;m&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;start&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;
        &lt;span class="k"&gt;if&lt;/span&gt; &lt;span class="n"&gt;gap&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;strip&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
            &lt;span class="n"&gt;errors&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;append&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;f&lt;/span&gt;&lt;span class="s2"&gt;"G2 line {cur_line}: stray text {gap.strip()[:30]!r}"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
        &lt;span class="n"&gt;cur_line&lt;/span&gt; &lt;span class="o"&gt;+=&lt;/span&gt; &lt;span class="n"&gt;gap&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;count&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="s2"&gt;"\n"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
        &lt;span class="n"&gt;line&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="n"&gt;cur_line&lt;/span&gt;
        &lt;span class="n"&gt;cur_line&lt;/span&gt; &lt;span class="o"&gt;+=&lt;/span&gt; &lt;span class="n"&gt;m&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;group&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="mi"&gt;0&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;count&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="s2"&gt;"\n"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
        &lt;span class="n"&gt;last_end&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="n"&gt;m&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;end&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;

        &lt;span class="n"&gt;is_close&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;tag&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;val&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="kt"&gt;bool&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;m&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;group&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="mi"&gt;1&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;m&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;group&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="mi"&gt;2&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;upper&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;m&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;group&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="mi"&gt;3&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;strip&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
        &lt;span class="k"&gt;if&lt;/span&gt; &lt;span class="ow"&gt;not&lt;/span&gt; &lt;span class="n"&gt;is_close&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
            &lt;span class="k"&gt;if&lt;/span&gt; &lt;span class="n"&gt;tag&lt;/span&gt; &lt;span class="k"&gt;in&lt;/span&gt; &lt;span class="no"&gt;CONTAINER_TAGS&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
                &lt;span class="n"&gt;last_leaf&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="kc"&gt;None&lt;/span&gt;
                &lt;span class="k"&gt;if&lt;/span&gt; &lt;span class="n"&gt;tag&lt;/span&gt; &lt;span class="o"&gt;==&lt;/span&gt; &lt;span class="s2"&gt;"STMTTRN"&lt;/span&gt; &lt;span class="ow"&gt;and&lt;/span&gt; &lt;span class="n"&gt;stack&lt;/span&gt; &lt;span class="ow"&gt;and&lt;/span&gt; &lt;span class="n"&gt;stack&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="o"&gt;-&lt;/span&gt;&lt;span class="mi"&gt;1&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="mi"&gt;0&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt; &lt;span class="o"&gt;==&lt;/span&gt; &lt;span class="s2"&gt;"STMTTRN"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
                    &lt;span class="n"&gt;_&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;prev_line&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="n"&gt;stack&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;pop&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
                    &lt;span class="n"&gt;errors&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;append&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;f&lt;/span&gt;&lt;span class="s2"&gt;"G2 line {line}: missing &amp;lt;/STMTTRN&amp;gt; opened at line {prev_line}"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
                    &lt;span class="k"&gt;if&lt;/span&gt; &lt;span class="n"&gt;current_tx&lt;/span&gt; &lt;span class="k"&gt;is&lt;/span&gt; &lt;span class="ow"&gt;not&lt;/span&gt; &lt;span class="kc"&gt;None&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
                        &lt;span class="n"&gt;txs&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;append&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;current_tx&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
                &lt;span class="n"&gt;stack&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;append&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;tag&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;line&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
                &lt;span class="k"&gt;if&lt;/span&gt; &lt;span class="n"&gt;tag&lt;/span&gt; &lt;span class="o"&gt;==&lt;/span&gt; &lt;span class="s2"&gt;"STMTTRN"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
                    &lt;span class="n"&gt;current_tx&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="s2"&gt;"line"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="n"&gt;line&lt;/span&gt;&lt;span class="p"&gt;}&lt;/span&gt;
            &lt;span class="k"&gt;else&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
                &lt;span class="n"&gt;last_leaf&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="n"&gt;tag&lt;/span&gt;
                &lt;span class="k"&gt;if&lt;/span&gt; &lt;span class="n"&gt;current_tx&lt;/span&gt; &lt;span class="k"&gt;is&lt;/span&gt; &lt;span class="ow"&gt;not&lt;/span&gt; &lt;span class="kc"&gt;None&lt;/span&gt; &lt;span class="ow"&gt;and&lt;/span&gt; &lt;span class="n"&gt;stack&lt;/span&gt; &lt;span class="ow"&gt;and&lt;/span&gt; &lt;span class="n"&gt;stack&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="o"&gt;-&lt;/span&gt;&lt;span class="mi"&gt;1&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="mi"&gt;0&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt; &lt;span class="o"&gt;==&lt;/span&gt; &lt;span class="s2"&gt;"STMTTRN"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
                    &lt;span class="n"&gt;current_tx&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="n"&gt;tag&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;current_tx&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="n"&gt;f&lt;/span&gt;&lt;span class="s2"&gt;"{tag}_line"&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="n"&gt;val&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;line&lt;/span&gt;
                &lt;span class="k"&gt;elif&lt;/span&gt; &lt;span class="n"&gt;stack&lt;/span&gt; &lt;span class="ow"&gt;and&lt;/span&gt; &lt;span class="n"&gt;stack&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="o"&gt;-&lt;/span&gt;&lt;span class="mi"&gt;1&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="mi"&gt;0&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt; &lt;span class="o"&gt;==&lt;/span&gt; &lt;span class="s2"&gt;"LEDGERBAL"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
                    &lt;span class="n"&gt;meta&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="n"&gt;f&lt;/span&gt;&lt;span class="s2"&gt;"LEDGERBAL_{tag}"&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="n"&gt;val&lt;/span&gt;
                &lt;span class="k"&gt;else&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
                    &lt;span class="n"&gt;meta&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="n"&gt;tag&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;meta&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="n"&gt;f&lt;/span&gt;&lt;span class="s2"&gt;"{tag}_line"&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="n"&gt;val&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="kt"&gt;str&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;line&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
        &lt;span class="k"&gt;elif&lt;/span&gt; &lt;span class="n"&gt;tag&lt;/span&gt; &lt;span class="k"&gt;in&lt;/span&gt; &lt;span class="no"&gt;CONTAINER_TAGS&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
            &lt;span class="n"&gt;last_leaf&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="kc"&gt;None&lt;/span&gt;
            &lt;span class="k"&gt;if&lt;/span&gt; &lt;span class="ow"&gt;not&lt;/span&gt; &lt;span class="n"&gt;stack&lt;/span&gt; &lt;span class="ow"&gt;or&lt;/span&gt; &lt;span class="n"&gt;stack&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="o"&gt;-&lt;/span&gt;&lt;span class="mi"&gt;1&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="mi"&gt;0&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt; &lt;span class="o"&gt;!=&lt;/span&gt; &lt;span class="n"&gt;tag&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
                &lt;span class="n"&gt;errors&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;append&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;f&lt;/span&gt;&lt;span class="s2"&gt;"G2 line {line}: unexpected &amp;lt;/{tag}&amp;gt;"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
                &lt;span class="k"&gt;while&lt;/span&gt; &lt;span class="n"&gt;stack&lt;/span&gt; &lt;span class="ow"&gt;and&lt;/span&gt; &lt;span class="n"&gt;stack&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="o"&gt;-&lt;/span&gt;&lt;span class="mi"&gt;1&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="mi"&gt;0&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt; &lt;span class="o"&gt;!=&lt;/span&gt; &lt;span class="n"&gt;tag&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
                    &lt;span class="n"&gt;unclosed&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;_&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="n"&gt;stack&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;pop&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
                    &lt;span class="k"&gt;if&lt;/span&gt; &lt;span class="n"&gt;unclosed&lt;/span&gt; &lt;span class="o"&gt;==&lt;/span&gt; &lt;span class="s2"&gt;"STMTTRN"&lt;/span&gt; &lt;span class="ow"&gt;and&lt;/span&gt; &lt;span class="n"&gt;current_tx&lt;/span&gt; &lt;span class="k"&gt;is&lt;/span&gt; &lt;span class="ow"&gt;not&lt;/span&gt; &lt;span class="kc"&gt;None&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
                        &lt;span class="n"&gt;txs&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;append&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;current_tx&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
                        &lt;span class="n"&gt;current_tx&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="kc"&gt;None&lt;/span&gt;
            &lt;span class="k"&gt;if&lt;/span&gt; &lt;span class="n"&gt;stack&lt;/span&gt; &lt;span class="ow"&gt;and&lt;/span&gt; &lt;span class="n"&gt;stack&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="o"&gt;-&lt;/span&gt;&lt;span class="mi"&gt;1&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="mi"&gt;0&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt; &lt;span class="o"&gt;==&lt;/span&gt; &lt;span class="n"&gt;tag&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
                &lt;span class="k"&gt;if&lt;/span&gt; &lt;span class="n"&gt;stack&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;pop&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="mi"&gt;0&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt; &lt;span class="o"&gt;==&lt;/span&gt; &lt;span class="s2"&gt;"STMTTRN"&lt;/span&gt; &lt;span class="ow"&gt;and&lt;/span&gt; &lt;span class="n"&gt;current_tx&lt;/span&gt; &lt;span class="k"&gt;is&lt;/span&gt; &lt;span class="ow"&gt;not&lt;/span&gt; &lt;span class="kc"&gt;None&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
                    &lt;span class="n"&gt;txs&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;append&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;current_tx&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
                    &lt;span class="n"&gt;current_tx&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="kc"&gt;None&lt;/span&gt;
        &lt;span class="k"&gt;elif&lt;/span&gt; &lt;span class="n"&gt;last_leaf&lt;/span&gt; &lt;span class="o"&gt;==&lt;/span&gt; &lt;span class="n"&gt;tag&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
            &lt;span class="n"&gt;last_leaf&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="kc"&gt;None&lt;/span&gt;

    &lt;span class="n"&gt;intu_bid&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="n"&gt;meta&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;get&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="s2"&gt;"INTU.BID"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;""&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
    &lt;span class="n"&gt;bid_line&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="n"&gt;meta&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;get&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="s2"&gt;"INTU.BID_line"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"1"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
    &lt;span class="k"&gt;if&lt;/span&gt; &lt;span class="ow"&gt;not&lt;/span&gt; &lt;span class="n"&gt;re&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;fullmatch&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;r&lt;/span&gt;&lt;span class="s2"&gt;"\d{1,10}"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;intu_bid&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="ow"&gt;or&lt;/span&gt; &lt;span class="kt"&gt;int&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;intu_bid&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="o"&gt;&amp;lt;=&lt;/span&gt; &lt;span class="mi"&gt;0&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
        &lt;span class="n"&gt;errors&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;append&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;f&lt;/span&gt;&lt;span class="s2"&gt;"G3 line {bid_line}: invalid INTU.BID={intu_bid!r}"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;

    &lt;span class="n"&gt;seen&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="kt"&gt;dict&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="kt"&gt;str&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="kt"&gt;int&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="p"&gt;}&lt;/span&gt;
    &lt;span class="n"&gt;deposits&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;withdrawals&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="nc"&gt;Decimal&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="s2"&gt;"0.00"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="nc"&gt;Decimal&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="s2"&gt;"0.00"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
    &lt;span class="k"&gt;for&lt;/span&gt; &lt;span class="n"&gt;tx&lt;/span&gt; &lt;span class="k"&gt;in&lt;/span&gt; &lt;span class="n"&gt;txs&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
        &lt;span class="n"&gt;line&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="kt"&gt;int&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;tx&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="s2"&gt;"line"&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
        &lt;span class="n"&gt;dt_line&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="kt"&gt;int&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;tx&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;get&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="s2"&gt;"DTPOSTED_line"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;line&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
        &lt;span class="k"&gt;if&lt;/span&gt; &lt;span class="nf"&gt;parse_ofx_datetime&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="kt"&gt;str&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;tx&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;get&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="s2"&gt;"DTPOSTED"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;""&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="k"&gt;is&lt;/span&gt; &lt;span class="kc"&gt;None&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
            &lt;span class="n"&gt;errors&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;append&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;f&lt;/span&gt;&lt;span class="s2"&gt;"G3 line {dt_line}: invalid DTPOSTED={tx.get('DTPOSTED')!r}"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
        &lt;span class="n"&gt;fitid&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="kt"&gt;str&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;tx&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;get&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="s2"&gt;"FITID"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;""&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
        &lt;span class="n"&gt;fid_line&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="kt"&gt;int&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;tx&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;get&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="s2"&gt;"FITID_line"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;line&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
        &lt;span class="k"&gt;if&lt;/span&gt; &lt;span class="ow"&gt;not&lt;/span&gt; &lt;span class="n"&gt;fitid&lt;/span&gt; &lt;span class="ow"&gt;or&lt;/span&gt; &lt;span class="nb"&gt;len&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;fitid&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="o"&gt;&amp;gt;&lt;/span&gt; &lt;span class="mi"&gt;255&lt;/span&gt; &lt;span class="ow"&gt;or&lt;/span&gt; &lt;span class="nb"&gt;any&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;c&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;isspace&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="k"&gt;for&lt;/span&gt; &lt;span class="n"&gt;c&lt;/span&gt; &lt;span class="k"&gt;in&lt;/span&gt; &lt;span class="n"&gt;fitid&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
            &lt;span class="n"&gt;errors&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;append&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;f&lt;/span&gt;&lt;span class="s2"&gt;"G4 line {fid_line}: malformed FITID={fitid!r}"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
        &lt;span class="k"&gt;elif&lt;/span&gt; &lt;span class="n"&gt;fitid&lt;/span&gt; &lt;span class="k"&gt;in&lt;/span&gt; &lt;span class="n"&gt;seen&lt;/span&gt; &lt;span class="ow"&gt;or&lt;/span&gt; &lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;known_fitids&lt;/span&gt; &lt;span class="ow"&gt;and&lt;/span&gt; &lt;span class="n"&gt;fitid&lt;/span&gt; &lt;span class="k"&gt;in&lt;/span&gt; &lt;span class="n"&gt;known_fitids&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
            &lt;span class="n"&gt;errors&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;append&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;f&lt;/span&gt;&lt;span class="s2"&gt;"G4 line {fid_line}: duplicate FITID={fitid!r} (first seen line {seen.get(fitid)})"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
        &lt;span class="n"&gt;seen&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;setdefault&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;fitid&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;fid_line&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;

        &lt;span class="n"&gt;amt&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="nc"&gt;Decimal&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="kt"&gt;str&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;tx&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="s2"&gt;"TRNAMT"&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
        &lt;span class="n"&gt;amt_line&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="kt"&gt;int&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;tx&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;get&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="s2"&gt;"TRNAMT_line"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;line&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
        &lt;span class="n"&gt;ttype&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="kt"&gt;str&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;tx&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;get&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="s2"&gt;"TRNTYPE"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;""&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;upper&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
        &lt;span class="k"&gt;if&lt;/span&gt; &lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;ttype&lt;/span&gt; &lt;span class="k"&gt;in&lt;/span&gt; &lt;span class="no"&gt;OUTFLOW_TYPES&lt;/span&gt; &lt;span class="ow"&gt;and&lt;/span&gt; &lt;span class="n"&gt;amt&lt;/span&gt; &lt;span class="o"&gt;&amp;gt;&lt;/span&gt; &lt;span class="mi"&gt;0&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="ow"&gt;or&lt;/span&gt; &lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;ttype&lt;/span&gt; &lt;span class="k"&gt;in&lt;/span&gt; &lt;span class="no"&gt;INFLOW_TYPES&lt;/span&gt; &lt;span class="ow"&gt;and&lt;/span&gt; &lt;span class="n"&gt;amt&lt;/span&gt; &lt;span class="o"&gt;&amp;lt;&lt;/span&gt; &lt;span class="mi"&gt;0&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
            &lt;span class="n"&gt;errors&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;append&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;f&lt;/span&gt;&lt;span class="s2"&gt;"G4 line {amt_line}: sign flip {ttype} with amount {amt}"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
        &lt;span class="n"&gt;deposits&lt;/span&gt; &lt;span class="o"&gt;+=&lt;/span&gt; &lt;span class="n"&gt;amt&lt;/span&gt; &lt;span class="k"&gt;if&lt;/span&gt; &lt;span class="n"&gt;amt&lt;/span&gt; &lt;span class="o"&gt;&amp;gt;=&lt;/span&gt; &lt;span class="mi"&gt;0&lt;/span&gt; &lt;span class="k"&gt;else&lt;/span&gt; &lt;span class="nc"&gt;Decimal&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="s2"&gt;"0.00"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
        &lt;span class="n"&gt;withdrawals&lt;/span&gt; &lt;span class="o"&gt;+=&lt;/span&gt; &lt;span class="o"&gt;-&lt;/span&gt;&lt;span class="n"&gt;amt&lt;/span&gt; &lt;span class="k"&gt;if&lt;/span&gt; &lt;span class="n"&gt;amt&lt;/span&gt; &lt;span class="o"&gt;&amp;lt;&lt;/span&gt; &lt;span class="mi"&gt;0&lt;/span&gt; &lt;span class="k"&gt;else&lt;/span&gt; &lt;span class="nc"&gt;Decimal&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="s2"&gt;"0.00"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;

    &lt;span class="n"&gt;diff&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="nc"&gt;Decimal&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;meta&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="s2"&gt;"LEDGERBAL_BALAMT"&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="o"&gt;-&lt;/span&gt; &lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;opening&lt;/span&gt; &lt;span class="o"&gt;+&lt;/span&gt; &lt;span class="n"&gt;deposits&lt;/span&gt; &lt;span class="o"&gt;-&lt;/span&gt; &lt;span class="n"&gt;withdrawals&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
    &lt;span class="k"&gt;if&lt;/span&gt; &lt;span class="n"&gt;diff&lt;/span&gt; &lt;span class="o"&gt;!=&lt;/span&gt; &lt;span class="nc"&gt;Decimal&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="s2"&gt;"0.00"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
        &lt;span class="n"&gt;errors&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;append&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;f&lt;/span&gt;&lt;span class="s2"&gt;"G5: balance mismatch diff={diff}"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
    &lt;span class="k"&gt;return&lt;/span&gt; &lt;span class="n"&gt;errors&lt;/span&gt;


&lt;span class="k"&gt;if&lt;/span&gt; &lt;span class="n"&gt;__name__&lt;/span&gt; &lt;span class="o"&gt;==&lt;/span&gt; &lt;span class="s2"&gt;"__main__"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
    &lt;span class="n"&gt;issues&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="nf"&gt;validate_ofx_file&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="nc"&gt;Path&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;sys&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="n"&gt;argv&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="mi"&gt;1&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="nc"&gt;Decimal&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;sys&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="n"&gt;argv&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="mi"&gt;2&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
    &lt;span class="k"&gt;print&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="s2"&gt;"PASS"&lt;/span&gt; &lt;span class="k"&gt;if&lt;/span&gt; &lt;span class="ow"&gt;not&lt;/span&gt; &lt;span class="n"&gt;issues&lt;/span&gt; &lt;span class="k"&gt;else&lt;/span&gt; &lt;span class="s2"&gt;"\n"&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;join&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;issues&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
    &lt;span class="n"&gt;sys&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;exit&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="mi"&gt;1&lt;/span&gt; &lt;span class="k"&gt;if&lt;/span&gt; &lt;span class="n"&gt;issues&lt;/span&gt; &lt;span class="k"&gt;else&lt;/span&gt; &lt;span class="mi"&gt;0&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Notice how &lt;code&gt;cur_line&lt;/code&gt; advances incrementally via &lt;code&gt;gap.count("\n")&lt;/code&gt; and &lt;code&gt;m.group(0).count("\n")&lt;/code&gt;. That single detail turns out to matter a lot at scale.&lt;/p&gt;

&lt;h2&gt;
  
  
  What six edge-case fixtures reveal across the five gates
&lt;/h2&gt;

&lt;p&gt;Testing the validator against &lt;code&gt;xml.etree.ElementTree.fromstring()&lt;/code&gt; across six synthetic OFX 1.02 checking statements (&lt;code&gt;10&lt;/code&gt; transactions each, &lt;code&gt;$4,250.00&lt;/code&gt; opening balance) shows why XML parsing gives a useless signal here.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F4xjfynism8ii50qz4box.webp" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F4xjfynism8ii50qz4box.webp" alt="Two-panel chart comparing xml.etree.ElementTree crashes against the five-gate script used to validate OFX 1.02 SGML FITID in Python" width="800" height="533"&gt;&lt;/a&gt;&lt;br&gt;
&lt;em&gt;Defects pinned per validation gate across six OFX 1.02 SGML fixtures (left) versus ElementTree parse crashes on every sample (right).&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;&lt;code&gt;ElementTree&lt;/code&gt; crashes at line 7 (&lt;code&gt;mismatched tag: line 7, column 2&lt;/code&gt;) on all six files because &lt;code&gt;&amp;lt;CODE&amp;gt;0&lt;/code&gt; lacks &lt;code&gt;&amp;lt;/CODE&amp;gt;&lt;/code&gt;, so it can't tell a valid bank download apart from a corrupted one. Our five-gate script clears the baseline statement and pins each defect to its exact rule:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;&lt;code&gt;01_valid_checking.ofx&lt;/code&gt; (&lt;code&gt;0&lt;/code&gt; issues)&lt;/strong&gt;: Parses &lt;code&gt;10&lt;/code&gt; transactions, accepts &lt;code&gt;INTU.BID=3000&lt;/code&gt;, confirms &lt;code&gt;10&lt;/code&gt; unique &lt;code&gt;FITID&lt;/code&gt; keys, and reconciles &lt;code&gt;$4,250.00 + $3,840.00 - $1,624.85 = $6,465.15&lt;/code&gt; (&lt;code&gt;diff=$0.00&lt;/code&gt;).&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;&lt;code&gt;02_duplicate_fitid.ofx&lt;/code&gt; (&lt;code&gt;1&lt;/code&gt; Gate 4 error)&lt;/strong&gt;: Transaction #8 reuses &lt;code&gt;20261009004&lt;/code&gt; from transaction #4 on line 69, catching the collision at line 97 before duplicate transactions enter matchers like &lt;a href="https://www.gnucash.org/docs/v5/C/gnucash-manual/trans-import.html" rel="noopener noreferrer"&gt;GnuCash's 42-day transaction importer&lt;/a&gt;.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;&lt;code&gt;03_missing_close_stmttrn.ofx&lt;/code&gt; (&lt;code&gt;1&lt;/code&gt; Gate 2 error)&lt;/strong&gt;: Dropping &lt;code&gt;&amp;lt;/STMTTRN&amp;gt;&lt;/code&gt; after transaction #3 triggers a container stack error at line 64 (&lt;code&gt;&amp;lt;STMTTRN&amp;gt;&lt;/code&gt; opened at line 58 was never closed) while keeping transaction #4 intact for downstream diagnostics.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;&lt;code&gt;04_bad_header_bid_date.ofx&lt;/code&gt; (&lt;code&gt;3&lt;/code&gt; errors in Gates 1 and 3)&lt;/strong&gt;: Flags a UTF-8 BOM ahead of &lt;code&gt;OFXHEADER:100&lt;/code&gt; at line 1, rejects non-numeric &lt;code&gt;&amp;lt;INTU.BID&amp;gt;BANK-09&lt;/code&gt; at line 24, and catches February 30 (&lt;code&gt;20260230093000&lt;/code&gt;) at line 67 where pattern-only regexes pass.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;&lt;code&gt;05_sign_flip_reversal.ofx&lt;/code&gt; (&lt;code&gt;11&lt;/code&gt; errors in Gates 4 and 5)&lt;/strong&gt;: Inverted CSV-to-OFX signs trigger &lt;code&gt;10&lt;/code&gt; polarity mismatches plus a &lt;code&gt;$4,430.30&lt;/code&gt; ledger gap.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;&lt;code&gt;06_ledger_mismatch.ofx&lt;/code&gt; (&lt;code&gt;1&lt;/code&gt; Gate 5 error)&lt;/strong&gt;: All SGML tags, dates, bank codes, and &lt;code&gt;FITID&lt;/code&gt; values validate cleanly, yet &lt;code&gt;&amp;lt;BALAMT&amp;gt;6322.65&lt;/code&gt; falls &lt;code&gt;-$142.50&lt;/code&gt; short of the calculated &lt;code&gt;$6,465.15&lt;/code&gt; close.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fzur5brxk00nuoeflg791.webp" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fzur5brxk00nuoeflg791.webp" alt="Line chart plotting cumulative Decimal running balances and LEDGERBAL mismatches across ten OFX 1.02 statement transactions" width="800" height="533"&gt;&lt;/a&gt;&lt;br&gt;
&lt;em&gt;Cumulative Decimal balance path for the clean checking statement ($6,465.15, $0.00 diff) against sign-flip reversal and a -$142.50 ledger mismatch.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;Look closely at &lt;code&gt;06_ledger_mismatch&lt;/code&gt;, because every line passes OFX 1.02 SGML grammar. Without computing &lt;code&gt;Opening + Deposits - Withdrawals&lt;/code&gt; in &lt;code&gt;Decimal&lt;/code&gt;, an importer posts that file and leaves a &lt;code&gt;$142.50&lt;/code&gt; discrepancy at close. Credit card statements (&lt;code&gt;CCSTMTRS&lt;/code&gt;) follow the same equation: purchases arrive as negative &lt;code&gt;&amp;lt;TRNAMT&amp;gt;&lt;/code&gt; values toward a negative &lt;code&gt;&amp;lt;LEDGERBAL&amp;gt;&amp;lt;BALAMT&amp;gt;&lt;/code&gt; liability, so keep the opening balance signed.&lt;/p&gt;

&lt;h2&gt;
  
  
  Throughput scaling across 100 to 10,000 transactions
&lt;/h2&gt;

&lt;p&gt;Benchmarking synthetic files from &lt;code&gt;100&lt;/code&gt; rows (&lt;code&gt;11.8 KB&lt;/code&gt;) up to &lt;code&gt;10,000&lt;/code&gt; rows (&lt;code&gt;1,115.9 KB&lt;/code&gt;) on Windows Python 3.13.3 exposed an &lt;code&gt;O(N^2)&lt;/code&gt; bug on my first pass.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Ffno6k018o1mx97xirb3g.webp" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Ffno6k018o1mx97xirb3g.webp" alt="Dual-panel benchmark chart showing linear O(N) stage runtime and log-scale speedup from 100 to 10,000 OFX 1.02 transactions in Python 3.13" width="800" height="533"&gt;&lt;/a&gt;&lt;br&gt;
&lt;em&gt;Median 15-run stage latency across 100 to 10,000 transactions (left) and log-scale speedup after fixing the O(N^2) line-count loop (right).&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;Initially I computed line numbers by calling &lt;code&gt;body_text.count("\n", 0, m.start())&lt;/code&gt; on each match. At &lt;code&gt;100&lt;/code&gt; rows (&lt;code&gt;6.10 ms&lt;/code&gt;) the bug hid; at &lt;code&gt;10,000&lt;/code&gt; rows rescanning from byte zero stalled for &lt;code&gt;33,699.44 ms&lt;/code&gt; (&lt;code&gt;297 tx/s&lt;/code&gt;). Replacing that with &lt;code&gt;cur_line += gap.count("\n")&lt;/code&gt; keeps every byte scanned once and cuts &lt;code&gt;10,000&lt;/code&gt; transactions (&lt;code&gt;15&lt;/code&gt;-run median) to &lt;code&gt;119.17 ms&lt;/code&gt;, or &lt;code&gt;83,914 tx/s&lt;/code&gt; (&lt;code&gt;76.25 ms&lt;/code&gt; in Gates 1-2, &lt;code&gt;42.92 ms&lt;/code&gt; in Gates 3-5). Across &lt;code&gt;100&lt;/code&gt; to &lt;code&gt;10,000&lt;/code&gt; rows, throughput holds steady between &lt;code&gt;74,074&lt;/code&gt; and &lt;code&gt;84,674 tx/s&lt;/code&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  When should you run this validator vs. ofxtools or ofxparse?
&lt;/h2&gt;

&lt;p&gt;Run this stdlib script as a preflight gate ahead of imports, rather than replacing &lt;code&gt;ofxtools&lt;/code&gt; on brokerage trees or signed Direct Connect HTTP calls:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Use this 5-gate stdlib validator when&lt;/strong&gt;: checking bank or credit-card OFX &lt;code&gt;1.02&lt;/code&gt;, &lt;code&gt;.qbo&lt;/code&gt;, or &lt;code&gt;.qfx&lt;/code&gt; exports before import, enforcing &lt;code&gt;&amp;lt;INTU.BID&amp;gt;&lt;/code&gt; numeric rules that &lt;code&gt;ofxtools&lt;/code&gt; strips, catching missing &lt;code&gt;&amp;lt;/STMTTRN&amp;gt;&lt;/code&gt; closures, and blocking duplicate &lt;code&gt;&amp;lt;FITID&amp;gt;&lt;/code&gt; keys or &lt;code&gt;Decimal&lt;/code&gt; balance gaps.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Use &lt;code&gt;ofxtools&lt;/code&gt; when&lt;/strong&gt;: parsing investment responses (&lt;code&gt;INVSTMTMSGSRSV1&lt;/code&gt;, &lt;code&gt;SECLIST&lt;/code&gt; security trees), handling OFX &lt;code&gt;2.x&lt;/code&gt; XML schemas, or generating bank requests.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Skip &lt;code&gt;xml.etree&lt;/code&gt; and regex tag-closers when&lt;/strong&gt;: bank exports mix &lt;code&gt;CP1252&lt;/code&gt; characters under &lt;code&gt;USASCII&lt;/code&gt; headers, omit &lt;code&gt;&amp;lt;/STMTTRN&amp;gt;&lt;/code&gt;, or prepend a &lt;code&gt;UTF-8&lt;/code&gt; BOM.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Pair the script with four downstream safeguards in production jobs:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Pass the prior reconciled &lt;code&gt;&amp;lt;LEDGERBAL&amp;gt;&amp;lt;BALAMT&amp;gt;&lt;/code&gt; as &lt;code&gt;opening&lt;/code&gt; so Gate 5 checks month-to-month continuity.&lt;/li&gt;
&lt;li&gt;Enforce a &lt;code&gt;UNIQUE&lt;/code&gt; &lt;code&gt;(account_id, fitid)&lt;/code&gt; index (&lt;code&gt;ON CONFLICT DO NOTHING&lt;/code&gt;) in SQLite or PostgreSQL across overlapping date windows.&lt;/li&gt;
&lt;li&gt;Decode raw bytes strictly via Gate 1 headers instead of calling &lt;code&gt;errors="ignore"&lt;/code&gt;.&lt;/li&gt;
&lt;li&gt;Quarantine batches raising &lt;code&gt;SIGN_FLIP_OUTFLOW_POSITIVE&lt;/code&gt; or &lt;code&gt;LEDGER_BALANCE_MISMATCH&lt;/code&gt; before posting to QuickBooks or GnuCash.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;&lt;em&gt;Editorial note: Portions of the outline and prose review for this tutorial were assisted by an LLM editor, while the Python 3.13 validator, synthetic OFX 1.02 test fixtures, scaling benchmarks, and matplotlib charts were written and executed locally by the author.&lt;/em&gt;&lt;/p&gt;

</description>
      <category>python</category>
      <category>fintech</category>
      <category>accounting</category>
      <category>tutorial</category>
    </item>
    <item>
      <title>AI character video pipeline in Python and FFmpeg for catching drift, word overflow, and bad crops</title>
      <dc:creator>Huyen Hai Nghi</dc:creator>
      <pubDate>Thu, 19 Nov 2026 17:16:07 +0000</pubDate>
      <link>https://dev.to/huyentoolnotes/ai-character-video-pipeline-in-python-and-ffmpeg-for-catching-drift-word-overflow-and-bad-crops-23nc</link>
      <guid>https://dev.to/huyentoolnotes/ai-character-video-pipeline-in-python-and-ffmpeg-for-catching-drift-word-overflow-and-bad-crops-23nc</guid>
      <description>&lt;p&gt;A reliable AI character video pipeline checks generated chunks for torso RGB drift, spoken word overflow, &lt;code&gt;-14 LUFS&lt;/code&gt; loudness swings, and unsafe crop boundaries before concatenating a single MP4. Skip those gates and three five-second clips stitched back to back will jump in color warmth, clip the character's mane on &lt;code&gt;16:9&lt;/code&gt; conversions, and desync dialogue mid-line.&lt;/p&gt;

&lt;p&gt;Most multi-shot AI tutorials stop right after running &lt;code&gt;ffmpeg -f concat&lt;/code&gt; on whatever files arrive from an inference queue. That works fine for scenic B-roll. When your recurring protagonist wears a tailored white suit, sunglasses, and gold jewelry across scenes, unvalidated concatenation ruins episode continuity.&lt;/p&gt;

&lt;h2&gt;
  
  
  Why an AI character video pipeline breaks between short clips
&lt;/h2&gt;

&lt;p&gt;Video diffusion backends render five-second to eight-second clips statelessly per request, so every cut in an AI character video pipeline drifts in exposure, color balance, gain, and framing unless local scripts gate intermediate files on disk. A dark convoy shot pulls midtones cool grey. A palace gate or golden throne cut pushes them amber.&lt;/p&gt;

&lt;p&gt;Emily Peng's walkthrough of the &lt;a href="https://www.kapwing.com/resources/how-to-make-sheru-bhai-ai-videos-prompts-tutorial/" rel="noopener noreferrer"&gt;Kapwing Sheru Bhai prompt workflow&lt;/a&gt; documents the same failure pattern across manual pipelines: white suiting melts into fur texture, and rushed Hindi lines break lip timing. Developers on Reddit automating Shorts hit that wall fast. Their practical fix is rendering five-second chunks to disk first, then forcing constant &lt;code&gt;fps&lt;/code&gt; and audio sample rates before stitching.&lt;/p&gt;

&lt;p&gt;You don't need custom validation plumbing if you work inside a preset studio like Sheru Bhai Maker, where character outfits, scene presets, and short spoken lines stay locked ahead of rendering. Once you orchestrate multi-provider batch jobs through Python scripts, though, automated checks between generation and assembly save real compute budget.&lt;/p&gt;

&lt;h2&gt;
  
  
  Measuring torso RGB and luma drift with Pillow
&lt;/h2&gt;

&lt;p&gt;Comparing full-frame perceptual hashes or global histograms between two scenes fails immediately because the background changes completely from an outdoor driveway to an indoor hall. Instead, extract a fixed torso region-of-interest (ROI) where the protagonist's outfit sits and measure RGB channel drift using &lt;a href="https://pillow.readthedocs.io/en/stable/reference/Image.html" rel="noopener noreferrer"&gt;Pillow's &lt;code&gt;Image&lt;/code&gt; module&lt;/a&gt; and &lt;code&gt;ImageStat&lt;/code&gt;.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Ft5a8qcihz1k3fk899hyy.webp" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Ft5a8qcihz1k3fk899hyy.webp" alt="Luminance curves and torso RGB channel bar chart measuring shot drift in an ai character video pipeline" width="800" height="533"&gt;&lt;/a&gt;&lt;br&gt;
&lt;em&gt;Luma density distributions (left) and torso-crop RGB channel means (right) expose warm amber vs cool neutral shifts across four scenes.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;The diagnostic chart above makes that exposure gap obvious. On the left, normalized pixel luminance curves show &lt;code&gt;Frame B (Boss Convoy)&lt;/code&gt; spiking near &lt;code&gt;15-25&lt;/code&gt; on dark tarmac, while &lt;code&gt;Frame A (Palace Gate)&lt;/code&gt; spreads higher into the &lt;code&gt;125-230&lt;/code&gt; range. On the right, torso RGB means separate four cuts cleanly: &lt;code&gt;Palace Gate&lt;/code&gt; leans warm amber (&lt;code&gt;R ~156, G ~126, B ~100&lt;/code&gt;), whereas &lt;code&gt;Boss Convoy&lt;/code&gt; sits cool grey (&lt;code&gt;R ~127, G ~128, B ~119&lt;/code&gt;). &lt;code&gt;Auction&lt;/code&gt; pushes higher under warm spotlight (&lt;code&gt;R ~176, G ~153, B ~130&lt;/code&gt;). Switch to &lt;code&gt;Golden Don&lt;/code&gt; and blue drops steeply (&lt;code&gt;R ~153, G ~109, B ~57&lt;/code&gt;) because of its metallic gold fabric.&lt;/p&gt;

&lt;p&gt;Notice why your pipeline must index &lt;code&gt;ref_rgb&lt;/code&gt; per wardrobe look rather than against one global constant. Comparing a &lt;code&gt;Golden Don&lt;/code&gt; shot against a &lt;code&gt;White Boss&lt;/code&gt; baseline will always trigger a false failure, whereas comparing two &lt;code&gt;White Boss&lt;/code&gt; cuts checks real lighting drift on the same white wool jacket. Computing weighted luminance (&lt;code&gt;0.299 * R + 0.587 * G + 0.114 * B&lt;/code&gt;) alongside &lt;code&gt;R - B&lt;/code&gt; warm skew catches both exposure collapse and color-temperature shifts in a single pass.&lt;/p&gt;

&lt;p&gt;I tested this drift checker on Python 3.13 against a &lt;code&gt;1500x1000&lt;/code&gt; sunset frame (&lt;code&gt;effect-wide-yacht.png&lt;/code&gt;). Sampling the torso ROI (&lt;code&gt;0.30..0.70&lt;/code&gt; x, &lt;code&gt;0.42..0.74&lt;/code&gt; y) against a reference RGB tuple &lt;code&gt;(210.0, 205.0, 195.0)&lt;/code&gt; returned &lt;code&gt;RGB = (168.9, 131.2, 98.4)&lt;/code&gt; and luminance &lt;code&gt;138.7&lt;/code&gt;. It also flagged a &lt;code&gt;max_channel_delta&lt;/code&gt; of &lt;code&gt;96.6&lt;/code&gt; and a warm skew (&lt;code&gt;R - B&lt;/code&gt;) of &lt;code&gt;+70.5&lt;/code&gt;.&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight python"&gt;&lt;code&gt;&lt;span class="k"&gt;from&lt;/span&gt; &lt;span class="no"&gt;PIL&lt;/span&gt; &lt;span class="k"&gt;import&lt;/span&gt; &lt;span class="nc"&gt;Image&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="nc"&gt;ImageStat&lt;/span&gt;


&lt;span class="k"&gt;def&lt;/span&gt; &lt;span class="nf"&gt;measure_torso_drift&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;frame_path&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="kt"&gt;str&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;ref_rgb&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="kt"&gt;tuple&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="kt"&gt;float&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="kt"&gt;float&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="kt"&gt;float&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="o"&gt;-&amp;gt;&lt;/span&gt; &lt;span class="kt"&gt;dict&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
    &lt;span class="s2"&gt;"""Measure RGB means, weighted luma, and warm skew inside a central torso ROI."""&lt;/span&gt;
    &lt;span class="k"&gt;with&lt;/span&gt; &lt;span class="nc"&gt;Image&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nb"&gt;open&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;frame_path&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="k"&gt;as&lt;/span&gt; &lt;span class="n"&gt;im&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
        &lt;span class="n"&gt;rgb&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="n"&gt;im&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;convert&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="s2"&gt;"RGB"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
        &lt;span class="n"&gt;w&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;h&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="n"&gt;rgb&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="n"&gt;size&lt;/span&gt;
        &lt;span class="n"&gt;torso_box&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="kt"&gt;int&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;w&lt;/span&gt; &lt;span class="o"&gt;*&lt;/span&gt; &lt;span class="mf"&gt;0.30&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="kt"&gt;int&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;h&lt;/span&gt; &lt;span class="o"&gt;*&lt;/span&gt; &lt;span class="mf"&gt;0.42&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="kt"&gt;int&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;w&lt;/span&gt; &lt;span class="o"&gt;*&lt;/span&gt; &lt;span class="mf"&gt;0.70&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="kt"&gt;int&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;h&lt;/span&gt; &lt;span class="o"&gt;*&lt;/span&gt; &lt;span class="mf"&gt;0.74&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
        &lt;span class="n"&gt;crop&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="n"&gt;rgb&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;crop&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;torso_box&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
        &lt;span class="n"&gt;r_mean&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;g_mean&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;b_mean&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="nc"&gt;ImageStat&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nc"&gt;Stat&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;crop&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="n"&gt;mean&lt;/span&gt;
        &lt;span class="n"&gt;luma&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="mf"&gt;0.299&lt;/span&gt; &lt;span class="o"&gt;*&lt;/span&gt; &lt;span class="n"&gt;r_mean&lt;/span&gt; &lt;span class="o"&gt;+&lt;/span&gt; &lt;span class="mf"&gt;0.587&lt;/span&gt; &lt;span class="o"&gt;*&lt;/span&gt; &lt;span class="n"&gt;g_mean&lt;/span&gt; &lt;span class="o"&gt;+&lt;/span&gt; &lt;span class="mf"&gt;0.114&lt;/span&gt; &lt;span class="o"&gt;*&lt;/span&gt; &lt;span class="n"&gt;b_mean&lt;/span&gt;
        &lt;span class="n"&gt;delta&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="nb"&gt;max&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;
            &lt;span class="nb"&gt;abs&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;r_mean&lt;/span&gt; &lt;span class="o"&gt;-&lt;/span&gt; &lt;span class="n"&gt;ref_rgb&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="mi"&gt;0&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
            &lt;span class="nb"&gt;abs&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;g_mean&lt;/span&gt; &lt;span class="o"&gt;-&lt;/span&gt; &lt;span class="n"&gt;ref_rgb&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="mi"&gt;1&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
            &lt;span class="nb"&gt;abs&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;b_mean&lt;/span&gt; &lt;span class="o"&gt;-&lt;/span&gt; &lt;span class="n"&gt;ref_rgb&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="mi"&gt;2&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
        &lt;span class="p"&gt;)&lt;/span&gt;
        &lt;span class="k"&gt;return&lt;/span&gt; &lt;span class="p"&gt;{&lt;/span&gt;
            &lt;span class="s2"&gt;"rgb_mean"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="nb"&gt;round&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;r_mean&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="mi"&gt;1&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="nb"&gt;round&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;g_mean&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="mi"&gt;1&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="nb"&gt;round&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;b_mean&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="mi"&gt;1&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
            &lt;span class="s2"&gt;"luma"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="nb"&gt;round&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;luma&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="mi"&gt;1&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
            &lt;span class="s2"&gt;"max_channel_delta"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="nb"&gt;round&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;delta&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="mi"&gt;1&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
            &lt;span class="s2"&gt;"warm_skew_r_minus_b"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="nb"&gt;round&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;r_mean&lt;/span&gt; &lt;span class="o"&gt;-&lt;/span&gt; &lt;span class="n"&gt;b_mean&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="mi"&gt;1&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
        &lt;span class="p"&gt;}&lt;/span&gt;


&lt;span class="k"&gt;def&lt;/span&gt; &lt;span class="nf"&gt;check_dialogue_budget&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;line&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="kt"&gt;str&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;clip_seconds&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="kt"&gt;int&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;max_words&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="kt"&gt;int&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="mi"&gt;10&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="o"&gt;-&amp;gt;&lt;/span&gt; &lt;span class="kt"&gt;tuple&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="kt"&gt;bool&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="kt"&gt;int&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="kt"&gt;str&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
    &lt;span class="s2"&gt;"""Reject spoken lines that exceed lip-sync pacing limits before rendering."""&lt;/span&gt;
    &lt;span class="n"&gt;words&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="n"&gt;w&lt;/span&gt; &lt;span class="k"&gt;for&lt;/span&gt; &lt;span class="n"&gt;w&lt;/span&gt; &lt;span class="k"&gt;in&lt;/span&gt; &lt;span class="n"&gt;line&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;strip&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;split&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="k"&gt;if&lt;/span&gt; &lt;span class="n"&gt;w&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;
    &lt;span class="n"&gt;limit&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="nb"&gt;min&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;max_words&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;clip_seconds&lt;/span&gt; &lt;span class="o"&gt;*&lt;/span&gt; &lt;span class="mi"&gt;2&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
    &lt;span class="k"&gt;if&lt;/span&gt; &lt;span class="nb"&gt;len&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;words&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="o"&gt;&amp;gt;&lt;/span&gt; &lt;span class="n"&gt;limit&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
        &lt;span class="k"&gt;return&lt;/span&gt; &lt;span class="kc"&gt;False&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="nb"&gt;len&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;words&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;f&lt;/span&gt;&lt;span class="s2"&gt;"Rejected: {len(words)} words &amp;gt; {limit} max for {clip_seconds}s"&lt;/span&gt;
    &lt;span class="k"&gt;return&lt;/span&gt; &lt;span class="kc"&gt;True&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="nb"&gt;len&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;words&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"OK"&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;You run this gate in plain Pillow without OpenCV binaries or GPU memory allocation. If &lt;code&gt;max_channel_delta&lt;/code&gt; trips your threshold across two shots wearing the same suit look, the script can halt the batch or send the clip to white-balance correction before muxing.&lt;/p&gt;

&lt;h2&gt;
  
  
  Guarding spoken word budgets and running FFmpeg loudnorm
&lt;/h2&gt;

&lt;p&gt;Lip-sync decoders break predictably when an upstream script feeds twenty-word sentences into a five-second slot. Running &lt;code&gt;check_dialogue_budget()&lt;/code&gt; early blocks bloated lines before TTS or video synthesis starts. Capping Roman Hindi, Hinglish, Urdu, or English dialogue strictly under 10 words (&lt;code&gt;~2&lt;/code&gt; words per second) preserves natural jaw timing across &lt;code&gt;5s&lt;/code&gt;, &lt;code&gt;8s&lt;/code&gt;, and &lt;code&gt;10s&lt;/code&gt; clips, and switching pure visual drops to music-only sub-bass avoids wasted dialogue tokens altogether.&lt;/p&gt;

&lt;p&gt;Even clean spoken clips still swing sharply in volume between heavy sub-bass intros and indoor voice lines. Single-pass dynamic gain pumps room noise upward during pauses. Instead, two-pass normalization with &lt;a href="https://ffmpeg.org/ffmpeg-filters.html" rel="noopener noreferrer"&gt;FFmpeg's &lt;code&gt;loudnorm&lt;/code&gt; filter&lt;/a&gt; measures &lt;code&gt;input_i&lt;/code&gt;, &lt;code&gt;input_tp&lt;/code&gt;, &lt;code&gt;input_lra&lt;/code&gt;, and &lt;code&gt;input_thresh&lt;/code&gt; first, then applies &lt;code&gt;linear=true&lt;/code&gt; scaling to land cleanly at &lt;code&gt;-14 LUFS&lt;/code&gt;. Setting &lt;code&gt;TP=-1.5&lt;/code&gt; leaves safe ceiling margin below &lt;code&gt;0.0 dBFS&lt;/code&gt; so AAC transcoding won't introduce inter-sample clipping, while locking &lt;code&gt;-ar 48000&lt;/code&gt; and &lt;code&gt;fps=30&lt;/code&gt; stops variable-frame-rate drift when mixed generator outputs are concatenated later.&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Pipeline Gate&lt;/th&gt;
&lt;th&gt;Target Parameter&lt;/th&gt;
&lt;th&gt;Tested Threshold / Setting&lt;/th&gt;
&lt;th&gt;Failure Mode Prevented&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Torso ROI RGB check&lt;/td&gt;
&lt;td&gt;Central crop &lt;code&gt;(0.30..0.70, 0.42..0.74)&lt;/code&gt;
&lt;/td&gt;
&lt;td&gt;Alert on &lt;code&gt;max_channel_delta &amp;gt; 45&lt;/code&gt; within matching look&lt;/td&gt;
&lt;td&gt;Unintended warm/cool white suit shift across cuts&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Dialogue word cap&lt;/td&gt;
&lt;td&gt;Spoken words per clip (&lt;code&gt;5s / 8s / 10s&lt;/code&gt;)&lt;/td&gt;
&lt;td&gt;Hard reject at &lt;code&gt;&amp;gt; min(10, seconds * 2)&lt;/code&gt; words&lt;/td&gt;
&lt;td&gt;Rushed syllables and desynced jaw animation&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Audio measurement (Pass 1)&lt;/td&gt;
&lt;td&gt;&lt;code&gt;loudnorm=I=-14:TP=-1.5:LRA=11:print_format=json&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;Parse JSON metrics from FFmpeg &lt;code&gt;stderr&lt;/code&gt;
&lt;/td&gt;
&lt;td&gt;Blind dynamic pumping on quiet dialogue pauses&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Linear export (Pass 2)&lt;/td&gt;
&lt;td&gt;
&lt;code&gt;linear=true&lt;/code&gt;, &lt;code&gt;-ar 48000&lt;/code&gt;, &lt;code&gt;fps=30&lt;/code&gt;
&lt;/td&gt;
&lt;td&gt;
&lt;code&gt;I=-14&lt;/code&gt; LUFS, &lt;code&gt;-1.5&lt;/code&gt; dBTP ceiling, &lt;code&gt;yuv420p&lt;/code&gt;
&lt;/td&gt;
&lt;td&gt;Volume jumps and VFR audio drift after concat&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Aspect crop bias&lt;/td&gt;
&lt;td&gt;Subject-anchored &lt;code&gt;x_bias&lt;/code&gt; / &lt;code&gt;y_bias&lt;/code&gt;
&lt;/td&gt;
&lt;td&gt;
&lt;code&gt;y_bias = 0.12&lt;/code&gt; (top-weighted) instead of &lt;code&gt;0.50&lt;/code&gt;
&lt;/td&gt;
&lt;td&gt;Cropping top of lion mane or off-center props&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;Call FFmpeg from Python using &lt;a href="https://docs.python.org/3/library/subprocess.html" rel="noopener noreferrer"&gt;&lt;code&gt;subprocess.run&lt;/code&gt; with &lt;code&gt;capture_output=True&lt;/code&gt; and &lt;code&gt;check=True&lt;/code&gt;&lt;/a&gt;. Capturing &lt;code&gt;stderr&lt;/code&gt; gives your parser direct access to the JSON block emitted by Pass 1, while &lt;code&gt;check=True&lt;/code&gt; raises &lt;code&gt;CalledProcessError&lt;/code&gt; immediately if a clip lacks an audio stream or fails decoding. If a silent music-less visual chunk enters this stage, skip &lt;code&gt;loudnorm&lt;/code&gt; or inject an &lt;code&gt;anullsrc&lt;/code&gt; track before normalization so the audio filtergraph doesn't abort.&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight python"&gt;&lt;code&gt;&lt;span class="k"&gt;import&lt;/span&gt; &lt;span class="n"&gt;json&lt;/span&gt;
&lt;span class="k"&gt;import&lt;/span&gt; &lt;span class="n"&gt;re&lt;/span&gt;
&lt;span class="k"&gt;import&lt;/span&gt; &lt;span class="n"&gt;subprocess&lt;/span&gt;


&lt;span class="k"&gt;def&lt;/span&gt; &lt;span class="nf"&gt;build_safe_crop_filter&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;
    &lt;span class="n"&gt;src_w&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="kt"&gt;int&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
    &lt;span class="n"&gt;src_h&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="kt"&gt;int&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
    &lt;span class="n"&gt;target_w&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="kt"&gt;int&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
    &lt;span class="n"&gt;target_h&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="kt"&gt;int&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
    &lt;span class="n"&gt;x_bias&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="kt"&gt;float&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="mf"&gt;0.5&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
    &lt;span class="n"&gt;y_bias&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="kt"&gt;float&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="mf"&gt;0.12&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="o"&gt;-&amp;gt;&lt;/span&gt; &lt;span class="kt"&gt;str&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
    &lt;span class="s2"&gt;"""Compute a head-safe crop and Lanczos scale filter without center-cropping out the crown."""&lt;/span&gt;
    &lt;span class="n"&gt;src_ratio&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="n"&gt;src_w&lt;/span&gt; &lt;span class="o"&gt;/&lt;/span&gt; &lt;span class="n"&gt;src_h&lt;/span&gt;
    &lt;span class="n"&gt;dst_ratio&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="n"&gt;target_w&lt;/span&gt; &lt;span class="o"&gt;/&lt;/span&gt; &lt;span class="n"&gt;target_h&lt;/span&gt;
    &lt;span class="k"&gt;if&lt;/span&gt; &lt;span class="n"&gt;src_ratio&lt;/span&gt; &lt;span class="o"&gt;&amp;gt;&lt;/span&gt; &lt;span class="n"&gt;dst_ratio&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
        &lt;span class="n"&gt;crop_h&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="n"&gt;src_h&lt;/span&gt;
        &lt;span class="n"&gt;crop_w&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="kt"&gt;int&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="nb"&gt;round&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;src_h&lt;/span&gt; &lt;span class="o"&gt;*&lt;/span&gt; &lt;span class="n"&gt;dst_ratio&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
        &lt;span class="n"&gt;x_off&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="kt"&gt;int&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="nb"&gt;round&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;src_w&lt;/span&gt; &lt;span class="o"&gt;-&lt;/span&gt; &lt;span class="n"&gt;crop_w&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="o"&gt;*&lt;/span&gt; &lt;span class="n"&gt;x_bias&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
        &lt;span class="n"&gt;y_off&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="mi"&gt;0&lt;/span&gt;
    &lt;span class="k"&gt;else&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
        &lt;span class="n"&gt;crop_w&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="n"&gt;src_w&lt;/span&gt;
        &lt;span class="n"&gt;crop_h&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="kt"&gt;int&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="nb"&gt;round&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;src_w&lt;/span&gt; &lt;span class="o"&gt;/&lt;/span&gt; &lt;span class="n"&gt;dst_ratio&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
        &lt;span class="n"&gt;x_off&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="mi"&gt;0&lt;/span&gt;
        &lt;span class="n"&gt;y_off&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="kt"&gt;int&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="nb"&gt;round&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;src_h&lt;/span&gt; &lt;span class="o"&gt;-&lt;/span&gt; &lt;span class="n"&gt;crop_h&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="o"&gt;*&lt;/span&gt; &lt;span class="n"&gt;y_bias&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
    &lt;span class="k"&gt;return&lt;/span&gt; &lt;span class="n"&gt;f&lt;/span&gt;&lt;span class="s2"&gt;"crop={crop_w}:{crop_h}:{x_off}:{y_off},scale={target_w}:{target_h}:flags=lanczos,setsar=1"&lt;/span&gt;


&lt;span class="k"&gt;def&lt;/span&gt; &lt;span class="nf"&gt;normalize_and_reframe&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;input_mp4&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="kt"&gt;str&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;output_mp4&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="kt"&gt;str&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;src_w&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="kt"&gt;int&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;src_h&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="kt"&gt;int&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="o"&gt;-&amp;gt;&lt;/span&gt; &lt;span class="kc"&gt;None&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
    &lt;span class="s2"&gt;"""Run two-pass EBU R128 loudness normalization (-14 LUFS) and safe 9:16 reframing."""&lt;/span&gt;
    &lt;span class="n"&gt;probe_cmd&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="p"&gt;[&lt;/span&gt;
        &lt;span class="s2"&gt;"ffmpeg"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"-hide_banner"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"-i"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;input_mp4&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
        &lt;span class="s2"&gt;"-af"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"loudnorm=I=-14:TP=-1.5:LRA=11:print_format=json"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
        &lt;span class="s2"&gt;"-f"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"null"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"-"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
    &lt;span class="p"&gt;]&lt;/span&gt;
    &lt;span class="n"&gt;probe&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="n"&gt;subprocess&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;run&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;probe_cmd&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;capture_output&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="kc"&gt;True&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;text&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="kc"&gt;True&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;check&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="kc"&gt;True&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
    &lt;span class="k"&gt;match&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="n"&gt;re&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;search&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;r&lt;/span&gt;&lt;span class="s2"&gt;"\{[^{}]*\"input_i\"[^{}]*\}"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;probe&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="n"&gt;stderr&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;re&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="no"&gt;S&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
    &lt;span class="k"&gt;if&lt;/span&gt; &lt;span class="ow"&gt;not&lt;/span&gt; &lt;span class="k"&gt;match&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
        &lt;span class="k"&gt;raise&lt;/span&gt; &lt;span class="nc"&gt;RuntimeError&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="s2"&gt;"Could not parse loudnorm JSON stats from FFmpeg stderr"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
    &lt;span class="n"&gt;stats&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="n"&gt;json&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;loads&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="k"&gt;match&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;group&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="mi"&gt;0&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;

    &lt;span class="n"&gt;vf&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="nf"&gt;build_safe_crop_filter&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;src_w&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;src_h&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="mi"&gt;1080&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="mi"&gt;1920&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;x_bias&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="mf"&gt;0.5&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;y_bias&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="mf"&gt;0.12&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
    &lt;span class="n"&gt;af&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="p"&gt;(&lt;/span&gt;
        &lt;span class="n"&gt;f&lt;/span&gt;&lt;span class="s2"&gt;"loudnorm=I=-14:TP=-1.5:LRA=11:"&lt;/span&gt;
        &lt;span class="n"&gt;f&lt;/span&gt;&lt;span class="s2"&gt;"measured_I={stats['input_i']}:measured_TP={stats['input_tp']}:"&lt;/span&gt;
        &lt;span class="n"&gt;f&lt;/span&gt;&lt;span class="s2"&gt;"measured_LRA={stats['input_lra']}:measured_thresh={stats['input_thresh']}:"&lt;/span&gt;
        &lt;span class="n"&gt;f&lt;/span&gt;&lt;span class="s2"&gt;"linear=true"&lt;/span&gt;
    &lt;span class="p"&gt;)&lt;/span&gt;
    &lt;span class="n"&gt;render_cmd&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="p"&gt;[&lt;/span&gt;
        &lt;span class="s2"&gt;"ffmpeg"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"-y"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"-i"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;input_mp4&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
        &lt;span class="s2"&gt;"-vf"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;vf&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"-af"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;af&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
        &lt;span class="s2"&gt;"-c:v"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"libx264"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"-pix_fmt"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"yuv420p"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"-r"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"30"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
        &lt;span class="s2"&gt;"-c:a"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"aac"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"-ar"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"48000"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"-b:a"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"192k"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
        &lt;span class="n"&gt;output_mp4&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
    &lt;span class="p"&gt;]&lt;/span&gt;
    &lt;span class="n"&gt;subprocess&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;run&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;render_cmd&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;capture_output&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="kc"&gt;True&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;text&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="kc"&gt;True&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;check&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="kc"&gt;True&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;I verified both passes locally under Python 3.13 and &lt;code&gt;ffmpeg 9.0.2-full_build-www.gyan.dev&lt;/code&gt; using a synthesized two-second &lt;code&gt;1500x1000&lt;/code&gt; H.264/AAC test clip (&lt;code&gt;effect-wide-yacht.png&lt;/code&gt; plus a 440 Hz sine wave). Pass 1 logged &lt;code&gt;input_i: -21.80&lt;/code&gt;, &lt;code&gt;input_tp: -17.65&lt;/code&gt;, &lt;code&gt;input_lra: 0.00&lt;/code&gt;, and &lt;code&gt;input_thresh: -31.80&lt;/code&gt;. Pass 2 produced a clean &lt;code&gt;175,411&lt;/code&gt;-byte &lt;code&gt;1080x1920&lt;/code&gt; &lt;code&gt;30 fps&lt;/code&gt; MP4.&lt;/p&gt;

&lt;h2&gt;
  
  
  Cross-cropping 9:16 and 16:9 frames without slicing the head
&lt;/h2&gt;

&lt;p&gt;Hardcoding FFmpeg's default center crop (&lt;code&gt;crop=ih*9/16:ih&lt;/code&gt; or &lt;code&gt;(ih-oh)/2&lt;/code&gt;) ruins framing whenever you repurpose &lt;code&gt;16:9&lt;/code&gt; widescreen masters for &lt;code&gt;9:16&lt;/code&gt; Shorts or vice versa. Simple ratio math shows why.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fbo4y9jszdef5t43aa5fn.webp" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fbo4y9jszdef5t43aa5fn.webp" alt="Bar chart showing pixel retention rates across 9:16, 4:5, 1:1, and 16:9 center crops in an ai character video pipeline" width="800" height="533"&gt;&lt;/a&gt;&lt;br&gt;
&lt;em&gt;Cross-cropping between 16:9 and 9:16 drops 68.4% of source pixels under center-crop rules, leaving only 31.6% of the frame.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;Center-cropping a &lt;code&gt;16:9&lt;/code&gt; master (&lt;code&gt;1920x1080&lt;/code&gt;) into &lt;code&gt;9:16&lt;/code&gt; (&lt;code&gt;1080x1920&lt;/code&gt;) discards &lt;code&gt;68.4%&lt;/code&gt; of the frame and retains just &lt;code&gt;31.6%&lt;/code&gt; of source pixels (&lt;code&gt;(9/16) / (16/9) = 81/256&lt;/code&gt;). Cropping &lt;code&gt;9:16&lt;/code&gt; down to a &lt;code&gt;4:5&lt;/code&gt; feed frame (&lt;code&gt;1080x1350&lt;/code&gt;) keeps &lt;code&gt;70.3%&lt;/code&gt;, &lt;code&gt;16:9&lt;/code&gt; to &lt;code&gt;4:5&lt;/code&gt; keeps &lt;code&gt;45.0%&lt;/code&gt;, and &lt;code&gt;1:1&lt;/code&gt; square holds &lt;code&gt;56.2%&lt;/code&gt; from either direction.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fcphnvrpzsqlkb6t71hgr.webp" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fcphnvrpzsqlkb6t71hgr.webp" alt="Wide sunset yacht helipad frame showing off-center lion character and hovering helicopter margin for aspect cropping" width="800" height="533"&gt;&lt;/a&gt;&lt;br&gt;
&lt;em&gt;On this 1500x1000 sunset frame, a naive 9:16 center crop cuts off the hovering helicopter beside the lion's shoulder.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;That &lt;code&gt;effect-wide-yacht.png&lt;/code&gt; still (&lt;code&gt;1500x1000&lt;/code&gt;, &lt;code&gt;3:2&lt;/code&gt;) shows the practical hazard cleanly. The lion boss stands slightly left of center while a hovering helicopter fills the upper right sky. A blind center crop (&lt;code&gt;x_bias=0.50&lt;/code&gt;) builds &lt;code&gt;crop=562:1000:469:0&lt;/code&gt; and amputates the helicopter rotor.&lt;/p&gt;

&lt;p&gt;Look at the vertical branch inside &lt;code&gt;build_safe_crop_filter()&lt;/code&gt; too. Cropping a tall &lt;code&gt;1080x1920&lt;/code&gt; portrait down to &lt;code&gt;16:9&lt;/code&gt; (&lt;code&gt;1080x608&lt;/code&gt;) under default center rules sets &lt;code&gt;y_off = 656&lt;/code&gt;, slicing straight through the lion's mane and aviators. Switching to &lt;code&gt;y_bias=0.12&lt;/code&gt; moves &lt;code&gt;y_off&lt;/code&gt; up to &lt;code&gt;157&lt;/code&gt; pixels from the top edge so vertical headroom stays intact. You can also shift &lt;code&gt;x_bias&lt;/code&gt; horizontally on wide frames to hold secondary props, though generating native aspect ratios beats discarding two-thirds of your pixels in post.&lt;/p&gt;

&lt;h2&gt;
  
  
  Where deterministic scripts stop paying off
&lt;/h2&gt;

&lt;p&gt;Pillow ROI checks and FFmpeg filters catch color shifts, dialogue overflow, gain spikes, and bad crop offsets. They stop dead on hallucinated anatomy. If a source render spawns bare lion paws under suit cuffs or extra fingers on a teacup, no histogram check or crop offset rescues those pixels downstream.&lt;/p&gt;

&lt;p&gt;Keep your scripts strictly in the gatekeeper role. Reject any segment whose torso RGB delta spikes past &lt;code&gt;45&lt;/code&gt; points on matching outfits, fail lines over &lt;code&gt;10&lt;/code&gt; words before audio synthesis, and run two-pass &lt;code&gt;-14 LUFS&lt;/code&gt; normalization only on validated cuts. Trying to repair broken source geometry in FFmpeg burns more hours than regenerating the bad shot.&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Editorial note: Portions of the outline and prose review for this tutorial were assisted by an LLM editor, while all Python code, FFmpeg invocations, aspect retention math, and local test runs were written and tested directly by the author.&lt;/em&gt;&lt;/p&gt;

</description>
      <category>tutorial</category>
    </item>
  </channel>
</rss>
