regexml 🚀

A lightweight, fault-tolerant, text-first XML extraction library for Python. Available on PyPI: pypi.org/project/regexml.

Unlike traditional XML parsers (xml.etree.ElementTree, lxml) that enforce strict document validation and crash with ParseError on incomplete or malformed markup, regexml treats XML purely as plain text. It extracts values directly using regular expressions and resilient pattern matching.

🌟 Why regexml?

• Fault-Tolerant & Resilient: Doesn't crash on unescaped &, <, >, missing closing tags, or broken XML headers.
• Perfect for LLM Outputs: Cleanly extract tags like <thought>, <answer>, or <code> from AI responses.
• Fluent Dot-Syntax Chaining: Call data.tagvalue("div").tagvalue("name") seamlessly.
• Balanced Outermost Tag Extraction: Accurately captures outer tags with nested tags of the same name without premature cut-offs.
• Truncated Stream Recovery: If an XML stream cuts off prematurely (e.g. at line 1000 of 1200), it safely captures data up to EOF without failing.
• Tag Inspector & Hierarchy: Built-in check_tags() discovers all tags, reports unclosed/truncated tags, and renders visual ASCII trees.
• Zero Heavy Dependencies: Built 100% on Python standard library.

📦 Installation

pip install regexml

⚡ Quick Start

1. Fluent Dot-Syntax (XMLText)

Wrap any string with XMLText to unlock dot-methods while retaining 100% string compatibility:

from regexml import XMLText

raw_data = """
<article id="101">
    <title>Resilient XML Extraction</title>
    <author>
        <first_name>Ishan</first_name>
        <last_name>Jain</last_name>
    </author>
</article>
"""

doc = XMLText(raw_data)

# Extract single tag value:
print(doc.tagvalue("title"))
# Output: Resilient XML Extraction

# Method Chaining:
print(doc.tagvalue("author").tagvalue("first_name"))
# Output: Ishan

# Path Shorthand:
print(doc.tagvalue("author/last_name"))
# Output: Jain

2. Functional API (Without Wrapping)

from regexml import tagvalue, tagvalues

data = "<user><name>Alice</name><name>Bob</name></user>"

print(tagvalue(data, "name"))        # Alice
print(tagvalues(data, "name"))       # ['Alice', 'Bob']

🎯 Advanced Features

Balanced Nested Tags of the Same Name

Standard regex chops off at the first closing tag. regexml tracks depth counter to return the true balanced outermost tag:

xml = """
<div id="outer">
    Hello
    <div id="inner">World</div>
    Welcome!
</div>
"""
doc = XMLText(xml)
print(doc.tagvalue("div"))
# Output:
# Hello
# <div id="inner">World</div>
# Welcome!

Truncated / Broken XML Recovery

If your XML was cut off mid-stream (e.g., from network drop or LLM token limit):

truncated_xml = "<user><id>123</id><bio>Software engineer who loves Python... [CUT OFF]"
doc = XMLText(truncated_xml)

bio = doc.tagvalue("bio")
print(bio)                  # "Software engineer who loves Python... [CUT OFF]"
print(bio.is_truncated)     # True

Tag Inspection & Broken Markup Detection (check_tags)

report = doc.check_tags()

# View all tags found:
print(report.tags)          # ['user', 'id', 'bio']

# Pinpoint unclosed tags (truncated points):
print(report.unclosed_tags) # ['user', 'bio']

# Print terminal ASCII tree:
print(report.tree())

Output:

└── user (unclosed)
    ├── id
    └── bio (unclosed)

The tree instantly visualizes the markup hierarchy and flags unclosed tags, making it easy to pinpoint where the XML stream cut off.

📄 License

MIT License. Developed by Ishan.