diffinite 0.8.0__tar.gz → 0.9.3__tar.gz

{diffinite-0.8.0/src/diffinite.egg-info → diffinite-0.9.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: diffinite
-Version: 0.8.0
+Version: 0.9.3
 Summary: Forensic source-code comparison tool — Winnowing fingerprints and professional PDF reports for IP litigation & code audit
 Author: nash-dir
 License: Apache-2.0
@@ -119,6 +119,8 @@ Each matched pair gets a side-by-side diff page with:
 
 - **Green highlight** — Lines present only in File B (additions)
 - **Red highlight** — Lines present only in File A (deletions)
+- **Purple highlight** — Lines moved from this position (`--detect-moved`)
+- **Blue highlight** — Lines moved to this position (`--detect-moved`)
 - **No highlight** — Identical lines (with configurable context folding)
 
 ### Deep Compare Section
@@ -139,7 +141,7 @@ Jaccard similarity is a well-defined set metric: `|A∩B| / |A∪B|`. Its interp
 |--------|-----------|----------|
 | `--page-number` | `Page 3 / 47` | Bottom-right |
 | `--file-number` | `File 2 / 12` | Bottom-left |
-| `--bates-number` | `DIFF-000003` | Bottom-center |
+| `--bates-number` | `TEST-000003-CONF` | Bottom-center |
 | `--show-filename` | `com/example/Foo.java` | Top-right |
 
 ---
@@ -178,6 +180,7 @@ dir_b    Path to the comparison source directory (B)
 | `--squash-blanks` | off | Collapse runs of 3+ blank lines. ⚠️ Changes line numbers — not recommended for forensic line-tracing. |
 | `--threshold N` | `60` | Fuzzy file-name matching threshold (0–100). Lower = more aggressive matching. |
 | `--collapse-identical` | off | Fold unchanged code blocks (3 context lines around each change) |
+| `--detect-moved` | off | Detect moved code blocks and highlight with distinct colors (purple=original, blue=destination) |
 
 ### Deep Compare Options
 
@@ -203,6 +206,9 @@ dir_b    Path to the comparison source directory (B)
 | `--page-number` | Show `Page n / N` at the bottom-right |
 | `--file-number` | Show `File n / N` at the bottom-left |
 | `--bates-number` | Stamp sequential Bates numbers at the bottom-center |
+| `--bates-prefix TEXT` | Bates number prefix (e.g. `PLAINTIFF-`). Combined as: `{prefix}{number}{suffix}` |
+| `--bates-suffix TEXT` | Bates number suffix (e.g. `-CONFIDENTIAL`) |
+| `--bates-start N` | Starting Bates number (default: `1`). Useful for continuing numbering across reports. |
 | `--show-filename` | Show filename at the top-right |
 
 ---
@@ -215,8 +221,9 @@ dir_b    Path to the comparison source directory (B)
 # Full forensic report with all annotations
 diffinite plaintiff_code/ defendant_code/ -o exhibit_A.pdf \
     --no-comments \
-    --bates-number --page-number --file-number --show-filename \
-    --collapse-identical
+    --bates-number --bates-prefix "CASE2026-" --bates-suffix "-CONFIDENTIAL" \
+    --bates-start 1 --page-number --file-number --show-filename \
+    --collapse-identical --detect-moved
 ```
 
 ### Code Audit (Quick HTML)

{diffinite-0.8.0 → diffinite-0.9.3}/README.md

@@ -82,6 +82,8 @@ Each matched pair gets a side-by-side diff page with:
 
 - **Green highlight** — Lines present only in File B (additions)
 - **Red highlight** — Lines present only in File A (deletions)
+- **Purple highlight** — Lines moved from this position (`--detect-moved`)
+- **Blue highlight** — Lines moved to this position (`--detect-moved`)
 - **No highlight** — Identical lines (with configurable context folding)
 
 ### Deep Compare Section
@@ -102,7 +104,7 @@ Jaccard similarity is a well-defined set metric: `|A∩B| / |A∪B|`. Its interp
 |--------|-----------|----------|
 | `--page-number` | `Page 3 / 47` | Bottom-right |
 | `--file-number` | `File 2 / 12` | Bottom-left |
-| `--bates-number` | `DIFF-000003` | Bottom-center |
+| `--bates-number` | `TEST-000003-CONF` | Bottom-center |
 | `--show-filename` | `com/example/Foo.java` | Top-right |
 
 ---
@@ -141,6 +143,7 @@ dir_b    Path to the comparison source directory (B)
 | `--squash-blanks` | off | Collapse runs of 3+ blank lines. ⚠️ Changes line numbers — not recommended for forensic line-tracing. |
 | `--threshold N` | `60` | Fuzzy file-name matching threshold (0–100). Lower = more aggressive matching. |
 | `--collapse-identical` | off | Fold unchanged code blocks (3 context lines around each change) |
+| `--detect-moved` | off | Detect moved code blocks and highlight with distinct colors (purple=original, blue=destination) |
 
 ### Deep Compare Options
 
@@ -166,6 +169,9 @@ dir_b    Path to the comparison source directory (B)
 | `--page-number` | Show `Page n / N` at the bottom-right |
 | `--file-number` | Show `File n / N` at the bottom-left |
 | `--bates-number` | Stamp sequential Bates numbers at the bottom-center |
+| `--bates-prefix TEXT` | Bates number prefix (e.g. `PLAINTIFF-`). Combined as: `{prefix}{number}{suffix}` |
+| `--bates-suffix TEXT` | Bates number suffix (e.g. `-CONFIDENTIAL`) |
+| `--bates-start N` | Starting Bates number (default: `1`). Useful for continuing numbering across reports. |
 | `--show-filename` | Show filename at the top-right |
 
 ---
@@ -178,8 +184,9 @@ dir_b    Path to the comparison source directory (B)
 # Full forensic report with all annotations
 diffinite plaintiff_code/ defendant_code/ -o exhibit_A.pdf \
     --no-comments \
-    --bates-number --page-number --file-number --show-filename \
-    --collapse-identical
+    --bates-number --bates-prefix "CASE2026-" --bates-suffix "-CONFIDENTIAL" \
+    --bates-start 1 --page-number --file-number --show-filename \
+    --collapse-identical --detect-moved
 ```
 
 ### Code Audit (Quick HTML)

{diffinite-0.8.0 → diffinite-0.9.3}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "diffinite"
-version = "0.8.0"
+version = "0.9.3"
 description = "Forensic source-code comparison tool — Winnowing fingerprints and professional PDF reports for IP litigation & code audit"
 readme = "README.md"
 license = {text = "Apache-2.0"}

diffinite-0.9.3/src/diffinite/__init__.py

@@ -0,0 +1,14 @@
+"""Diffinite — Forensic source-code diff tool.
+
+Compare two source directories, track logic movement across files
+(N:M cross-matching), and generate syntax-highlighted PDF reports.
+"""
+
+from importlib.metadata import version as _pkg_version, PackageNotFoundError
+
+try:
+    __version__: str = _pkg_version("diffinite")
+except PackageNotFoundError:
+    # Fallback for editable installs or running from source without install
+    __version__ = "0.0.0-dev"
+

{diffinite-0.8.0 → diffinite-0.9.3}/src/diffinite/cli.py

@@ -95,6 +95,34 @@ def main(argv: list[str] | None = None) -> None:
         default=FUZZY_THRESHOLD,
         help=f"Fuzzy matching threshold 0–100 (default: {FUZZY_THRESHOLD})",
     )
+    parser.add_argument(
+        "--encoding",
+        default="auto",
+        help=(
+            "Source file encoding. 'auto' (default) uses charset-normalizer "
+            "auto-detection with Korean-optimized fallback (utf-8 -> euc-kr -> cp949). "
+            "Specify an explicit encoding (e.g. euc-kr, utf-8, cp949, shift_jis, "
+            "gb2312) to force-decode all files with that encoding."
+        ),
+    )
+    parser.add_argument(
+        "--sort-by",
+        choices=["filename", "size", "ratio"],
+        default=None,
+        dest="sort_by",
+        help=(
+            "Sort matched file pairs in the report. "
+            "'filename' sorts by file path, 'size' by file size, "
+            "'ratio' by similarity ratio. Default: insertion order (no sort)."
+        ),
+    )
+    parser.add_argument(
+        "--sort-order",
+        choices=["asc", "desc"],
+        default="asc",
+        dest="sort_order",
+        help="Sort direction (default: asc). Only effective with --sort-by.",
+    )
 
     # ── Output modes ──────────────────────────────────────────────────
     parser.add_argument(
@@ -123,6 +151,33 @@ def main(argv: list[str] | None = None) -> None:
         default=False,
         help="Stamp Bates numbers at the bottom-center of each page",
     )
+    parser.add_argument(
+        "--bates-prefix",
+        type=str,
+        default="",
+        help=(
+            "Bates number prefix (e.g. 'PLAINTIFF-'). "
+            "Combined as: {prefix}{number}{suffix}"
+        ),
+    )
+    parser.add_argument(
+        "--bates-suffix",
+        type=str,
+        default="",
+        help=(
+            "Bates number suffix (e.g. '-CONFIDENTIAL'). "
+            "Combined as: {prefix}{number}{suffix}"
+        ),
+    )
+    parser.add_argument(
+        "--bates-start",
+        type=int,
+        default=1,
+        help=(
+            "Starting Bates number (default: 1). "
+            "Useful for continuing numbering across multiple reports."
+        ),
+    )
     parser.add_argument(
         "--show-filename",
         action="store_true",
@@ -139,6 +194,16 @@ def main(argv: list[str] | None = None) -> None:
             "Without this flag, the full diff is shown."
         ),
     )
+    parser.add_argument(
+        "--detect-moved",
+        action="store_true",
+        default=False,
+        help=(
+            "Detect moved code blocks and highlight them with distinct colors "
+            "(purple=original position, blue=moved position) instead of "
+            "plain delete/add. Works in both simple and deep modes."
+        ),
+    )
 
     # ── Report format options ─────────────────────────────────────────
     format_group = parser.add_argument_group(
@@ -280,6 +345,9 @@ def main(argv: list[str] | None = None) -> None:
         autojunk=not args.no_autojunk,
     )
 
+    # Resolve encoding
+    encoding = args.encoding if args.encoding.lower() != "auto" else None
+
     run_pipeline(
         dir_a=args.dir_a,
         dir_b=args.dir_b,
@@ -313,6 +381,17 @@ def main(argv: list[str] | None = None) -> None:
         report_html=args.report_html,
         report_md=args.report_md,
         report_json=args.report_json,
+        # Encoding
+        encoding=encoding,
+        # Sorting
+        sort_by=args.sort_by,
+        sort_order=args.sort_order,
+        # Moved block detection
+        detect_moved=args.detect_moved,
+        # Bates prefix/suffix
+        bates_prefix=args.bates_prefix,
+        bates_suffix=args.bates_suffix,
+        bates_start=args.bates_start,
     )