summaryrefslogtreecommitdiffstats
path: root/contrib/python/markdown-it-py/markdown_it/common/html_re.py
diff options
context:
space:
mode:
authoreivanov89 <[email protected]>2025-08-29 10:12:02 +0300
committereivanov89 <[email protected]>2025-08-29 10:27:27 +0300
commit140ced4d34c422c9f3cbe096f8dd35243b67d6e4 (patch)
treeb7373341f64151c0ab9839ee692dc919366590d5 /contrib/python/markdown-it-py/markdown_it/common/html_re.py
parent136471c8b2f3ab8cd7993200c0de0456b7018118 (diff)
Add python/textual to YDB
commit_hash:eda16a869229724fec5479fa27fa5cdbccbe0395
Diffstat (limited to 'contrib/python/markdown-it-py/markdown_it/common/html_re.py')
-rw-r--r--contrib/python/markdown-it-py/markdown_it/common/html_re.py39
1 files changed, 39 insertions, 0 deletions
diff --git a/contrib/python/markdown-it-py/markdown_it/common/html_re.py b/contrib/python/markdown-it-py/markdown_it/common/html_re.py
new file mode 100644
index 00000000000..ab822c5fc48
--- /dev/null
+++ b/contrib/python/markdown-it-py/markdown_it/common/html_re.py
@@ -0,0 +1,39 @@
+"""Regexps to match html elements"""
+
+import re
+
+attr_name = "[a-zA-Z_:][a-zA-Z0-9:._-]*"
+
+unquoted = "[^\"'=<>`\\x00-\\x20]+"
+single_quoted = "'[^']*'"
+double_quoted = '"[^"]*"'
+
+attr_value = "(?:" + unquoted + "|" + single_quoted + "|" + double_quoted + ")"
+
+attribute = "(?:\\s+" + attr_name + "(?:\\s*=\\s*" + attr_value + ")?)"
+
+open_tag = "<[A-Za-z][A-Za-z0-9\\-]*" + attribute + "*\\s*\\/?>"
+
+close_tag = "<\\/[A-Za-z][A-Za-z0-9\\-]*\\s*>"
+comment = "<!---?>|<!--(?:[^-]|-[^-]|--[^>])*-->"
+processing = "<[?][\\s\\S]*?[?]>"
+declaration = "<![A-Za-z][^>]*>"
+cdata = "<!\\[CDATA\\[[\\s\\S]*?\\]\\]>"
+
+HTML_TAG_RE = re.compile(
+ "^(?:"
+ + open_tag
+ + "|"
+ + close_tag
+ + "|"
+ + comment
+ + "|"
+ + processing
+ + "|"
+ + declaration
+ + "|"
+ + cdata
+ + ")"
+)
+HTML_OPEN_CLOSE_TAG_STR = "^(?:" + open_tag + "|" + close_tag + ")"
+HTML_OPEN_CLOSE_TAG_RE = re.compile(HTML_OPEN_CLOSE_TAG_STR)