wowdespacito/anthropics_skills
1
0
Fork 0
mirror of https://github.com/anthropics/skills.git synced 2026-08-02 13:05:28 +08:00
Code Issues Packages Projects Releases Wiki Activity

Update docx, pptx, and xlsx skills (#1447)

Browse Source 
Add support for template formats (.dotx, .potx, .xltx) across validation
and the helper scripts.

Consolidate the shared office helpers into a single module, replacing the
pack/unpack pipeline with explicit zip/unzip steps. Extraction now rejects
symlink and path-traversal archive entries. Move run merging to a
standalone docx script, since it only ever applied to Word documents.

Fix the redlining validator so it compares against the original even when
a document has no tracked changes, which is when an untracked edit would
otherwise go unreported.

Provision a LibreOffice user profile per invocation so conversions work in
sandboxed environments.

Trim the skill docs to the guidance that earns its place.
This commit is contained in:
Peter Lai
2026-07-16 19:47:37 -07:00
committed by GitHub
parent 9d2f1ae187
commit fa0fa64bdc
53 changed files with 4147 additions and 4442 deletions
Download Patch File Download Diff File Expand all files Collapse all files
skills/xlsx/scripts/office/validators/redlining.py
+96 -44
View File
@@ -1,5 +1,14 @@
"""
Validator for tracked changes in Word documents.
Detects untracked edits in word/document.xml: text that differs from the
original without a <w:ins>/<w:del> wrapper recording it. The tracked changes
that are new relative to the original are undone, and the result is compared
against the original; whatever text still differs was edited without being
tracked.
Only the document body is compared. Headers, footers, footnotes and endnotes
are separate parts and are not checked.
"""
import subprocess
@@ -7,14 +16,18 @@ import tempfile
import zipfile
from pathlib import Path
import defusedxml.ElementTree as ET
from defusedxml.common import DefusedXmlException
from helpers import rendered_text, safe_extract
class RedliningValidator:
def __init__(self, unpacked_dir, original_docx, verbose=False, author="Claude"):
def __init__(self, unpacked_dir, original_docx, verbose=False):
self.unpacked_dir = Path(unpacked_dir)
self.original_docx = Path(original_docx)
self.verbose = verbose
self.author = author
self.namespaces = {
"w": "http://schemas.openxmlformats.org/wordprocessingml/2006/main"
}
@@ -28,40 +41,12 @@ class RedliningValidator:
print(f"FAILED - Modified document.xml not found at {modified_file}")
return False
try:
import xml.etree.ElementTree as ET
tree = ET.parse(modified_file)
root = tree.getroot()
del_elements = root.findall(".//w:del", self.namespaces)
ins_elements = root.findall(".//w:ins", self.namespaces)
author_del_elements = [
elem
for elem in del_elements
if elem.get(f"{{{self.namespaces['w']}}}author") == self.author
]
author_ins_elements = [
elem
for elem in ins_elements
if elem.get(f"{{{self.namespaces['w']}}}author") == self.author
]
if not author_del_elements and not author_ins_elements:
if self.verbose:
print(f"PASSED - No tracked changes by {self.author} found.")
return True
except Exception:
pass
with tempfile.TemporaryDirectory() as temp_dir:
temp_path = Path(temp_dir)
try:
with zipfile.ZipFile(self.original_docx, "r") as zip_ref:
zip_ref.extractall(temp_path)
safe_extract(zip_ref, temp_path)
except Exception as e:
print(f"FAILED - Error unpacking original docx: {e}")
return False
@@ -74,18 +59,16 @@ class RedliningValidator:
return False
try:
import xml.etree.ElementTree as ET
modified_tree = ET.parse(modified_file)
modified_root = modified_tree.getroot()
original_tree = ET.parse(original_file)
original_root = original_tree.getroot()
except ET.ParseError as e:
except (ET.ParseError, DefusedXmlException) as e:
print(f"FAILED - Error parsing XML files: {e}")
return False
self._remove_author_tracked_changes(original_root)
self._remove_author_tracked_changes(modified_root)
new_changes = self._new_tracked_changes(original_root, modified_root)
self._remove_tracked_changes(modified_root, new_changes)
modified_text = self._extract_text_content(modified_root)
original_text = self._extract_text_content(original_root)
@@ -98,20 +81,91 @@ class RedliningValidator:
return False
if self.verbose:
print(f"PASSED - All changes by {self.author} are properly tracked")
print(
f"PASSED - All {len(new_changes)} change(s) against the original "
"are properly tracked"
)
return True
def _tracked_change_elements(self, root):
ins_tag = f"{{{self.namespaces['w']}}}ins"
del_tag = f"{{{self.namespaces['w']}}}del"
return [elem for elem in root.iter() if elem.tag in (ins_tag, del_tag)]
def _rendered_text(self, elem):
preserve = elem.get("{http://www.w3.org/XML/1998/namespace}space") == "preserve"
return rendered_text(elem.text or "", preserve)
def _text_elements(self, elem):
w = self.namespaces["w"]
return [
node
for node in elem.iter()
if node.tag in (f"{{{w}}}t", f"{{{w}}}delText")
]
def _tracked_change_key(self, elem):
w = self.namespaces["w"]
text = "".join(self._rendered_text(node) for node in self._text_elements(elem))
return (elem.tag, elem.get(f"{{{w}}}author"), elem.get(f"{{{w}}}date"), text)
def _new_tracked_changes(self, original_root, modified_root):
original = self._tracked_change_elements(original_root)
modified = self._tracked_change_elements(modified_root)
pool = {}
for elem in original:
pool.setdefault(self._tracked_change_key(elem), []).append(elem)
matched, leftover = set(), []
for elem in modified:
bucket = pool.get(self._tracked_change_key(elem))
if bucket:
matched.add(bucket.pop())
else:
leftover.append(elem)
def group(elem):
return self._tracked_change_key(elem)[:3]
def text_of(elems):
return "".join(self._tracked_change_key(e)[3] for e in elems)
unmatched_original = {}
for elem in original:
if elem not in matched:
unmatched_original.setdefault(group(elem), []).append(elem)
by_group = {}
for elem in leftover:
by_group.setdefault(group(elem), []).append(elem)
new = set()
for key, elems in by_group.items():
rebuilt = text_of(elems)
if rebuilt and rebuilt == text_of(unmatched_original.get(key, [])):
continue
new.update(elems)
return new
def _generate_detailed_diff(self, original_text, modified_text):
error_parts = [
f"FAILED - Document text doesn't match after removing {self.author}'s tracked changes",
"FAILED - Document text doesn't match after removing the tracked changes",
"",
"Likely causes:",
" 1. Modified text inside another author's <w:ins> or <w:del> tags",
" 2. Made edits without proper tracked changes",
" 3. Didn't nest <w:del> inside <w:ins> when deleting another's insertion",
" 4. Rewrote another author's <w:ins>/<w:del> and changed its text on",
" the way. A tracked change from the original is recognised by its",
" author, date and text; anything that doesn't reproduce one exactly",
" reads as new, and the text it carried is reported missing.",
"",
"For pre-redlined documents, use correct patterns:",
" - To reject another's INSERTION: Nest <w:del> inside their <w:ins>",
" - To reject PART of one: nest <w:del> around only the runs you reject.",
" Their <w:ins> may be split around it, so long as the pieces keep",
" their author and date and still spell out the same text.",
" - To restore another's DELETION: Add new <w:ins> AFTER their <w:del>",
"",
]
@@ -195,15 +249,14 @@ class RedliningValidator:
return None
def _remove_author_tracked_changes(self, root):
def _remove_tracked_changes(self, root, targets):
ins_tag = f"{{{self.namespaces['w']}}}ins"
del_tag = f"{{{self.namespaces['w']}}}del"
author_attr = f"{{{self.namespaces['w']}}}author"
for parent in root.iter():
to_remove = []
for child in parent:
if child.tag == ins_tag and child.get(author_attr) == self.author:
if child.tag == ins_tag and child in targets:
to_remove.append(child)
for elem in to_remove:
parent.remove(elem)
@@ -214,7 +267,7 @@ class RedliningValidator:
for parent in root.iter():
to_process = []
for child in parent:
if child.tag == del_tag and child.get(author_attr) == self.author:
if child.tag == del_tag and child in targets:
to_process.append((child, list(parent).index(child)))
for del_elem, del_index in reversed(to_process):
@@ -234,8 +287,7 @@ class RedliningValidator:
for p_elem in root.findall(f".//{p_tag}"):
text_parts = []
for t_elem in p_elem.findall(f".//{t_tag}"):
if t_elem.text:
text_parts.append(t_elem.text)
text_parts.append(self._rendered_text(t_elem))
paragraph_text = "".join(text_parts)
if paragraph_text:
paragraphs.append(paragraph_text)