From 29f0a282a8efb9ea93d33670ad42f1c151f70243 Mon Sep 17 00:00:00 2001
From: jesopo
Date: Sun, 9 Feb 2020 10:49:10 +0000
Subject: add utils.parse.sed.

---
 src/utils/parse/__init__.py |  3 +-
 src/utils/parse/sed.py      | 73 +++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 75 insertions(+), 1 deletion(-)
 create mode 100644 src/utils/parse/sed.py

(limited to 'src')

diff --git a/src/utils/parse/__init__.py b/src/utils/parse/__init__.py
index cf32d8f5..dd13ecbc 100644
--- a/src/utils/parse/__init__.py
+++ b/src/utils/parse/__init__.py
@@ -1,9 +1,10 @@
-import decimal, io, typing
+import decimal, io, re, typing
 from src.utils import datetime, errors
 
 from .spec import *
 from .time import duration
 from .types import try_int
+from . import sed
 
 COMMENT_TYPES = ["#", "//"]
 def hashflags(filename: str
diff --git a/src/utils/parse/sed.py b/src/utils/parse/sed.py
new file mode 100644
index 00000000..96234063
--- /dev/null
+++ b/src/utils/parse/sed.py
@@ -0,0 +1,73 @@
+import dataclasses, re, typing
+
+class Sed(object):
+    type: str
+    def match(self, s: str) -> typing.Optional[str]:
+        return None
+
+@dataclasses.dataclass
+class SedReplace(Sed):
+    type: str
+    pattern: typing.Pattern
+    replace: str
+    count: int
+
+    def match(self, s):
+        return self.pattern.sub(self.replace, s, self.count)
+
+@dataclasses.dataclass
+class SedMatch(Sed):
+    type: str
+    pattern: typing.Pattern
+
+    def match(self, s):
+        match = self.pattern.search(s)
+        if match:
+            return match.group(0)
+        return None
+
+def _sed_split(s):
+    backslash = False
+    forward_slash = []
+    for i, c in enumerate(s):
+        if not backslash:
+            if c == "/":
+                forward_slash.append(i)
+            if c == "\\":
+                backslash = True
+        else:
+            backslash = False
+    if forward_slash and (not forward_slash[-1] == (len(s)-1)):
+        forward_slash.append(len(s))
+
+    last = 0
+    out = []
+    for i in forward_slash:
+        out.append(s[last:i])
+        last = i+1
+    return out
+
+def _sed_flags(s: str) -> typing.Tuple[int, int]:
+    count = 1
+    re_flags = 0
+    if "g" in s:
+        count = 0
+    if "i" in s:
+        re_flags |= re.I
+    return count, re_flags
+
+def parse_sed(sed_s: str) -> typing.Optional[Sed]:
+    type, pattern, *args = _sed_split(sed_s)
+    if type == "s":
+        replace, *args = args
+        count, flags = _sed_flags((args or [""])[0])
+        pattern = re.compile(pattern, flags)
+        return SedReplace(type, pattern, replace, count)
+    elif type == "m":
+        count, flags = _sed_flags((args or [""])[0])
+        return SedMatch(type, re.compile(pattern, flags))
+    return None
+
+def sed(sed_obj: Sed, s: str) -> typing.Tuple[str, typing.Optional[str]]:
+    out = sed_obj.match(s)
+    return sed_obj.type, out
-- 
cgit v1.3.1-10-gc9f91