ignoring tex-math when math:mml alternative exists (#109)

* ignoring tex-math when math-:mmll alternative exists * changes to base.py * made tex-math fix viable for iop as well * lint fix * minor formatting fix * lint fix * minor edit to decompose all tex elements --------- Co-authored-by: Mugdha Polimera <[email protected]> Co-authored-by: Mugdha Polimera <[email protected]>
adsabs · May 29, 2024 · 9dd796b · 9dd796b
1 parent 6e254d1
commit 9dd796b
Show file tree

Hide file tree

Showing 11 changed files with 14,920 additions and 7 deletions.
diff --git a/adsingestp/parsers/base.py b/adsingestp/parsers/base.py
@@ -543,6 +543,14 @@ def _detag(self, r, tags_keep):
             for e in elements:
                 if t in self.HTML_TAGS_DANGER:
                     e.decompose()
+                elif (t == "alternatives") or (t == "inline-formula"):
+                    alt_math_element = e.find_all("mml:math", [])
+                    alt_tex_element = e.find_all("tex-math", [])
+                    if alt_math_element and alt_tex_element:
+                        for ee in alt_tex_element:
+                            ee.decompose()
+                    if t not in tags_keep:
+                        e.unwrap()
                 elif t in tags_keep:
                     continue
                 else:

diff --git a/tests/stubdata/input/jats_apj_967_1_35.xml b/tests/stubdata/input/jats_apj_967_1_35.xml