1 files changed, 146 insertions, 0 deletions
diff --git a/kodereviewer/diff_parser.py b/kodereviewer/diff_parser.py
new file mode 100644
index 0000000..378c795
--- /dev/null
+++ b/kodereviewer/diff_parser.py
@@ -0,0 +1,146 @@
+import logging
+import re
+from collections.abc import Sequence
+
+from rich.logging import RichHandler
+
+HEADER_RE = re.compile(r'@@ -(?P<original_start_line>\d+),(?P<original_line_count>\d+)'
+                       r' \+(?P<new_start_line>\d+),(?P<new_line_count>\d+)'
+                       r' @@ ?(?P<context>.*)')
+
+logging.basicConfig(
+    level="INFO", format="%(message)s", datefmt="[%X]", handlers=[RichHandler()]
+)
+logger = logging.getLogger(__name__)
+
+
+class SideBySideDiffLine:
+    line: int
+    original: str
+    new: str
+
+    def __init__(self, line: int, original: str, new: str):
+        self.line = line
+        self.original = original
+        self.new = new
+
+
+class Hunk:
+    header: str
+    lines: list[str]
+
+    original_start_line: int
+    original_line_count: int
+    new_start_line: int
+    new_line_count: int
+    context: str
+
+    original: list[str]
+    new: list[str]
+
+    def __init__(self, hunk_str: str):
+        header, *body = hunk_str.splitlines()
+        self.header = header
+        self.lines = body
+
+        self.parse_header(header)
+        self.parse_body(body)
+
+    def parse_header(self, header: str) -> None:
+        """Parses a header.
+        The format is: @@ -l,s +l,s @@ optional section heading
+        where
+          l: starting line
+          s: number of lines
+
+        example: @@ -13,10 +13,15 @@ on:
+        """
+        match = HEADER_RE.match(header)
+        if match:
+            self.original_start_line = int(match.group('original_start_line'))
+            self.original_line_count = int(match.group('original_line_count'))
+            self.new_start_line = int(match.group('new_start_line'))
+            self.new_line_count = int(match.group('new_line_count'))
+            self.context = match.group('context')
+        else:
+            raise ValueError(f'Error parsing {header}')
+
+    def parse_body(self, lines: list[str]) -> None:
+        self.original = []
+        self.new = []
+
+        for line in lines:
+            if line.startswith('-'):
+                self.original.append(line)
+            elif line.startswith('+'):
+                self.new.append(line)
+            else:
+                self.original.append(line)
+                self.new.append(line)
+
+    def side_by_side(self) -> Sequence[SideBySideDiffLine]:
+        start = min(self.original_start_line, self.new_start_line)
+        end = max(self.original_start_line + self.original_line_count, self.new_start_line, self.new_line_count)
+        retval: list[SideBySideDiffLine] = []
+        original_line_counter = 0
+        new_line_counter = 0
+
+        for line in range(start, end):
+            original = ''
+            new = ''
+            if line in range(self.original_start_line, self.original_start_line + self.original_line_count):
+                original = self.original[original_line_counter]
+                original_line_counter += 1
+
+            if line in range(self.new_start_line, self.new_start_line + self.new_line_count):
+                new = self.new[new_line_counter]
+                new_line_counter += 1
+
+            retval.append(SideBySideDiffLine(line, original, new))
+
+        return retval
+
+
+class GithubDiffParser:
+    diff: str
+
+    def __init__(self, diff: str):
+        self.diff = diff
+
+    def hunks(self) -> list[Hunk]:
+        retval: list[Hunk] = []
+
+        hunk_str = ''
+        for line in self.diff.splitlines():
+            if line.startswith('@@'):
+                logger.info('adding %s', line)
+                if hunk_str != '':
+                    retval.append(Hunk(hunk_str))
+                hunk_str = ''
+            hunk_str += line
+            hunk_str += '\n'
+        if hunk_str != '':
+            retval.append(Hunk(hunk_str))
+        return retval
+
+if __name__ == '__main__':
+    from rich.console import Console
+    from rich.table import Table
+
+    import sys
+    filename = sys.argv[-1]
+
+    with open(filename) as fp:
+        parser = GithubDiffParser(fp.read())
+        for hunk in parser.hunks():
+
+            table = Table(title=hunk.context)
+            table.add_column('line')
+            table.add_column('original')
+            table.add_column('new')
+
+            for diff_line in hunk.side_by_side():
+                table.add_row(str(diff_line.line), diff_line.original, diff_line.new)
+
+            console = Console()
+            console.print(table)