Issue #20 Initial implementation of zip listing support

duviz.py

@@ -53,6 +53,10 @@ class SubprocessException(RuntimeError):
     pass
 
 
+class ParseException(ValueError):
+    pass
+
+
 class SizeTree:
     """
     Base class for a tree of nodes where each node has a size and zero or more sub-nodes.
@@ -210,6 +214,51 @@ def pairs(listing: str) -> Iterator[Tuple[List[str], int]]:
         return tree
 
 
+class ZipListingParseException(ParseException):
+    pass
+
+
+def size_tree_from_zip_listing(listing: Iterable[str]) -> SizeTree:
+    def pairs(listing: Iterator[str]) -> Iterator[Tuple[List[str], int]]:
+        row_regex = re.compile(r"^\s*(\d+)\s+[\d-]+\s+[\d:]+\s+(.*)$")
+        for line in listing:
+            if line.startswith("---"):
+                # Reached table end
+                return
+            mo = row_regex.match(line)
+            if not mo:
+                raise ZipListingParseException(
+                    f"Failed to parse zip listing line {line}"
+                )
+            size, path = mo.group(1, 2)
+            yield path_split(path), int(size)
+
+    listing = iter(listing)
+    # Try to parse zip listing header
+    archive_line = next(listing)
+    mo = re.match(r"^Archive:\s*(.+)$", archive_line)
+    if not mo:
+        raise ZipListingParseException(
+            f"Failed to parse Archive name from {archive_line}."
+        )
+    archive_name = mo.group(1)
+
+    header_line = next(listing)
+    table_header = re.split(r"\s+", header_line.strip())
+    if table_header != ["Length", "Date", "Time", "Name"]:
+        raise ZipListingParseException(f"Unexpected table header {header_line}.")
+    header_separator = next(listing)
+    header_separator_regex = re.compile(r"^[- ]+$")
+    if not header_separator_regex.match(header_separator):
+        raise ZipListingParseException(
+            f"Unexpected table separator {header_separator}."
+        )
+
+    tree = SizeTree.from_path_size_pairs(pairs=pairs(listing), root=archive_name)
+    tree._recalculate_own_sizes_to_total_sizes()
+    return tree
+
+
 class SizeFormatter:
     """Render a (byte) count in compact human-readable way: 12, 34k, 56M, ..."""
 

test_duviz.py

@@ -5,9 +5,23 @@
 
 import pytest
 
-from duviz import TreeRenderer, SIZE_FORMATTER_COUNT, SIZE_FORMATTER_BYTES, SIZE_FORMATTER_BYTES_BINARY, path_split, \
-    SizeTree, AsciiDoubleLineBarRenderer, DuTree, InodeTree, get_progress_reporter, AsciiSingleLineBarRenderer, \
-    ColorDoubleLineBarRenderer, ColorSingleLineBarRenderer, Colorizer
+from duviz import (
+    SIZE_FORMATTER_BYTES,
+    SIZE_FORMATTER_BYTES_BINARY,
+    SIZE_FORMATTER_COUNT,
+    AsciiDoubleLineBarRenderer,
+    AsciiSingleLineBarRenderer,
+    ColorDoubleLineBarRenderer,
+    Colorizer,
+    ColorSingleLineBarRenderer,
+    DuTree,
+    InodeTree,
+    SizeTree,
+    TreeRenderer,
+    get_progress_reporter,
+    path_split,
+    size_tree_from_zip_listing,
+)
 
 
 def test_bar_one():
@@ -644,3 +658,39 @@ def test_get_progress_reporter():
     deltas = [i1-i0 for (i0, i1) in zip(indexes[:-1], indexes[1:])]
     assert all(d < 5 for d in deltas[:5])
     assert all(d > 9 for d in deltas[-5:])
+
+
+class TestZipListing:
+    def _check_render(self, listing: str, expected: str, width=40):
+        tree = size_tree_from_zip_listing(_dedent(listing).split("\n"))
+        result = AsciiDoubleLineBarRenderer().render(tree, width=width)
+        assert result == _dedent_and_split(expected)
+
+    def test_basic(self):
+        self._check_render(
+            listing="""
+                Archive:  tmp.zip
+                  Length      Date    Time    Name
+                ---------  ---------- -----   ----
+                        0  2022-10-08 00:27   docs/
+                        0  2022-10-08 00:27   docs/img/
+                     5456  2022-10-08 00:27   docs/img/logo.png
+                     3634  2022-10-08 00:26   docs/api.txt
+                     1575  2022-10-08 00:26   docs/intro.txt
+                     2244  2022-10-08 00:25   README.md
+                ---------                     -------
+                    12909                     6 files
+                """,
+            expected="""
+                ________________________________________________________________________________
+                [                                   tmp.zip                                    ]
+                [____________________________________12.91k____________________________________]
+                [                              docs                              ][ README.md  ]
+                [_____________________________10.66k_____________________________][___2.24k____]
+                [              img              ][       api.txt       ][intro.tx]              
+                [_____________5.46k_____________][________3.63k________][_1.57k__]              
+                [            logo.png           ]                                               
+                [_____________5.46k_____________]                                               
+                """,
+            width=80,
+        )