Re-apply: Move glob_to_regex and re_word_boundary to matrix-python-co…

…mmon #11505 (#11687)

Co-authored-by: Sean Quah <seanq@element.io>

changelog.d/11505.misc

@@ -0,0 +1 @@
+Move `glob_to_regex` and `re_word_boundary` to `matrix-python-common`.

changelog.d/11687.misc

@@ -0,0 +1 @@
+Move `glob_to_regex` and `re_word_boundary` to `matrix-python-common`.

synapse/config/room_directory.py

@@ -15,8 +15,9 @@
 
 from typing import List
 
+from matrix_common.regex import glob_to_regex
+
 from synapse.types import JsonDict
-from synapse.util import glob_to_regex
 
 from ._base import Config, ConfigError
 

synapse/config/tls.py

@@ -16,11 +16,12 @@
 import os
 from typing import List, Optional, Pattern
 
+from matrix_common.regex import glob_to_regex
+
 from OpenSSL import SSL, crypto
 from twisted.internet._sslverify import Certificate, trustRootFromCertificates
 
 from synapse.config._base import Config, ConfigError
-from synapse.util import glob_to_regex
 
 logger = logging.getLogger(__name__)
 

synapse/federation/federation_server.py

@@ -28,6 +28,7 @@
     Union,
 )
 
+from matrix_common.regex import glob_to_regex
 from prometheus_client import Counter, Gauge, Histogram
 
 from twisted.internet.abstract import isIPAddress
@@ -65,7 +66,7 @@
 )
 from synapse.storage.databases.main.lock import Lock
 from synapse.types import JsonDict, get_domain_from_id
-from synapse.util import glob_to_regex, json_decoder, unwrapFirstError
+from synapse.util import json_decoder, unwrapFirstError
 from synapse.util.async_helpers import Linearizer, concurrently_execute, gather_results
 from synapse.util.caches.response_cache import ResponseCache
 from synapse.util.stringutils import parse_server_name

synapse/push/push_rule_evaluator.py

@@ -17,9 +17,10 @@
 import re
 from typing import Any, Dict, List, Optional, Pattern, Tuple, Union
 
+from matrix_common.regex import glob_to_regex, to_word_pattern
+
 from synapse.events import EventBase
 from synapse.types import JsonDict, UserID
-from synapse.util import glob_to_regex, re_word_boundary
 from synapse.util.caches.lrucache import LruCache
 
 logger = logging.getLogger(__name__)
@@ -184,7 +185,7 @@ def _contains_display_name(self, display_name: Optional[str]) -> bool:
         r = regex_cache.get((display_name, False, True), None)
         if not r:
             r1 = re.escape(display_name)
-            r1 = re_word_boundary(r1)
+            r1 = to_word_pattern(r1)
             r = re.compile(r1, flags=re.IGNORECASE)
             regex_cache[(display_name, False, True)] = r
 
@@ -213,7 +214,7 @@ def _glob_matches(glob: str, value: str, word_boundary: bool = False) -> bool:
     try:
         r = regex_cache.get((glob, True, word_boundary), None)
         if not r:
-            r = glob_to_regex(glob, word_boundary)
+            r = glob_to_regex(glob, word_boundary=word_boundary)
             regex_cache[(glob, True, word_boundary)] = r
         return bool(r.search(value))
     except re.error:

synapse/python_dependencies.py

@@ -88,6 +88,7 @@
     # with the latest security patches.
     "cryptography>=3.4.7",
     "ijson>=3.1",
+    "matrix-common==1.0.0",
 ]
 
 CONDITIONAL_REQUIREMENTS = {

synapse/util/__init__.py

@@ -14,9 +14,8 @@
 
 import json
 import logging
-import re
 import typing
-from typing import Any, Callable, Dict, Generator, Optional, Pattern
+from typing import Any, Callable, Dict, Generator, Optional
 
 import attr
 from frozendict import frozendict
@@ -35,9 +34,6 @@
 logger = logging.getLogger(__name__)
 
 
-_WILDCARD_RUN = re.compile(r"([\?\*]+)")
-
-
 def _reject_invalid_json(val: Any) -> None:
     """Do not allow Infinity, -Infinity, or NaN values in JSON."""
     raise ValueError("Invalid JSON value: '%s'" % val)
@@ -185,56 +181,3 @@ def log_failure(
     if not consumeErrors:
         return failure
     return None
-
-
-def glob_to_regex(glob: str, word_boundary: bool = False) -> Pattern:
-    """Converts a glob to a compiled regex object.
-
-    Args:
-        glob: pattern to match
-        word_boundary: If True, the pattern will be allowed to match at word boundaries
-           anywhere in the string. Otherwise, the pattern is anchored at the start and
-           end of the string.
-
-    Returns:
-        compiled regex pattern
-    """
-
-    # Patterns with wildcards must be simplified to avoid performance cliffs
-    # - The glob `?**?**?` is equivalent to the glob `???*`
-    # - The glob `???*` is equivalent to the regex `.{3,}`
-    chunks = []
-    for chunk in _WILDCARD_RUN.split(glob):
-        # No wildcards? re.escape()
-        if not _WILDCARD_RUN.match(chunk):
-            chunks.append(re.escape(chunk))
-            continue
-
-        # Wildcards? Simplify.
-        qmarks = chunk.count("?")
-        if "*" in chunk:
-            chunks.append(".{%d,}" % qmarks)
-        else:
-            chunks.append(".{%d}" % qmarks)
-
-    res = "".join(chunks)
-
-    if word_boundary:
-        res = re_word_boundary(res)
-    else:
-        # \A anchors at start of string, \Z at end of string
-        res = r"\A" + res + r"\Z"
-
-    return re.compile(res, re.IGNORECASE)
-
-
-def re_word_boundary(r: str) -> str:
-    """
-    Adds word boundary characters to the start and end of an
-    expression to require that the match occur as a whole word,
-    but do so respecting the fact that strings starting or ending
-    with non-word characters will change word boundaries.
-    """
-    # we can't use \b as it chokes on unicode. however \W seems to be okay
-    # as shorthand for [^0-9A-Za-z_].
-    return r"(^|\W)%s(\W|$)" % (r,)