snippet translate: fix get_snippets

- Fixed the get_snippets function which did not work properly
  when more than one snippet id was on the same line.
- Tests were added

Pick-to: 6.2 6.3
Change-Id: Idffbb0aee258522d7855e2ad0e2b8df61a1872c8
Reviewed-by: Friedemann Kleint <Friedemann.Kleint@qt.io>

3 files changed, 128 insertions, 25 deletions

diff --git a/tools/snippets_translate/main.py b/tools/snippets_translate/main.py
index 5e31f50ef..cf34bc6cd 100644
--- a/tools/snippets_translate/main.py
+++ b/tools/snippets_translate/main.py
@@ -9,6 +9,7 @@ import sys
 from enum import Enum
 from pathlib import Path
 from textwrap import dedent
+from typing import List
 
 from converter import snippet_translate
 
@@ -149,28 +150,44 @@ def is_valid_file(x):
     return True
 
 
-def get_snippet_ids(line):
-    """Extract the snippet ids for a line '//! [1] //! [2]'"""
+def get_snippet_ids(line: str) -> List[str]:
+    # Extract the snippet ids for a line '//! [1] //! [2]'
     result = []
     for m in SNIPPET_PATTERN.finditer(line):
         result.append(m.group(1))
     return result
 
 
-def get_snippets(data):
-    """Extract (potentially overlapping) snippets from a C++ file indicated by //! [1]"""
-    current_snippets = []  # Active ids
-    snippets = []
-    for line in data:
-        new_ids = get_snippet_ids(line)
-        for id in new_ids:
-            if id in current_snippets:  # id encountered 2nd time: Snippet ends
-                current_snippets.remove(id)
-            else:
-                current_snippets.append(id)
-
-        if new_ids or current_snippets:
-            snippets.append(line)
+def get_snippets(lines: List[str]) -> List[List[str]]:
+    # Extract (potentially overlapping) snippets from a C++ file indicated by //! [1]
+    snippets: List[List[str]] = []
+    snippet: List[str]
+
+    i = 0
+    while i < len(lines):
+        line = lines[i]
+        i += 1
+
+        start_ids = get_snippet_ids(line)
+        while start_ids:
+            # Start of a snippet
+            start_id = start_ids.pop(0)
+            snippet = [line]  # The snippet starts with his id
+
+            # Find the end of the snippet
+            j = i
+            while j < len(lines):
+                l = lines[j]
+                j += 1
+
+                # Add the line to the snippet
+                snippet.append(l)
+
+                # Check if the snippet is complete
+                if start_id in get_snippet_ids(l):
+                    # End of snippet
+                    snippets.append(snippet)
+                    break
 
     return snippets
 
diff --git a/tools/snippets_translate/tests/test_converter.py b/tools/snippets_translate/tests/test_converter.py
index 17ab0b449..be46e0c0b 100644
--- a/tools/snippets_translate/tests/test_converter.py
+++ b/tools/snippets_translate/tests/test_converter.py
@@ -355,10 +355,12 @@ def test_ternary_operator():
         == "if not game.saveGame(json if Game.Json else Game.Binary):"
     )
 
+
 def test_useless_qt_classes():
     assert st('result += QLatin1String("; ");') == 'result += "; "'
     assert st("<< QLatin1Char('\0') << endl;") == "print('\0')"
 
+
 def test_special_cases():
     assert (
         st('http->setProxy("proxy.example.com", 3128);')
diff --git a/tools/snippets_translate/tests/test_snippets.py b/tools/snippets_translate/tests/test_snippets.py
index 6977b8278..6c451c84e 100644
--- a/tools/snippets_translate/tests/test_snippets.py
+++ b/tools/snippets_translate/tests/test_snippets.py
@@ -1,17 +1,101 @@
 # Copyright (C) 2022 The Qt Company Ltd.
 # SPDX-License-Identifier: LicenseRef-Qt-Commercial OR LGPL-3.0-only OR GPL-2.0-only OR GPL-3.0-only
 
-from main import get_snippets
+from main import get_snippets, get_snippet_ids
 
 
-SNIPPETS = ["pretext",
-            "//![some name]", "line1",
-            "//! [some name] [some other name]",
-            "line2",
-            "//! [some other name]",
-            "posttext"]
+def test_stacking():
+    lines = [
+        "//! [A] //! [B] ",
+        "//! [C] //! [D] //! [E]",
+        "// Content",
+        "//! [C] //! [A] ",
+        "//! [B] //! [D] //! [E]",
+    ]
+    snippets = get_snippets(lines)
+    assert len(snippets) == 5
+    assert len(snippets[0]) == 4  # A starts at line 0 and ends at line 3
+    assert len(snippets[1]) == 5  # B starts at line 0 and ends at line 4
+    assert len(snippets[2]) == 3  # C starts at line 1 and ends at line 3
+    assert len(snippets[3]) == 4  # D starts at line 1 and ends at line 4
+    assert len(snippets[4]) == 4  # E starts at line 1 and ends at line 4
+
+
+def test_nesting():
+    lines = [
+        "//! [A]",
+        "//! [B]",
+        "//! [C]",
+        "// Content",
+        "//! [A]",
+        "//! [C]",
+        "//! [B]",
+    ]
+    snippets = get_snippets(lines)
+    assert len(snippets) == 3
+
+    assert len(snippets[0]) == 5
+    assert snippets[0] == lines[:5]
+
+    assert len(snippets[1]) == 6
+    assert snippets[1] == lines[1:]
+
+    assert len(snippets[2]) == 4
+    assert snippets[2] == lines[2:6]
+
+
+def test_overlapping():
+    lines = [
+        "pretext",
+        "//! [A]",
+        "l1",
+        "//! [C]",
+        "//! [A] //! [B]",
+        "l2",
+        "l3 // Comment",
+        "//! [B]",
+        "posttext",
+        "//! [C]",
+    ]
+    snippets = get_snippets(lines)
+    assert len(snippets) == 3
+
+    assert len(snippets[0]) == 4
+    assert snippets[0] == lines[1:5]
+
+    assert len(snippets[1]) == 7
+    assert snippets[1] == lines[3:]
+
+    assert len(snippets[2]) == 4
+    assert snippets[2] == lines[4:8]
 
 
 def test_snippets():
-    extracted = get_snippets(SNIPPETS)
-    assert len(extracted) == len(SNIPPETS) - 2
+    lines = [
+        "pretext",
+        "//! [A]",
+        "l1",
+        "//! [A] //! [B]",
+        "l2",
+        "l3 // Comment",
+        "//! [B]",
+        "posttext"
+    ]
+
+    snippets = get_snippets(lines)
+    assert len(snippets) == 2
+
+    assert len(snippets[0]) == 3
+    assert snippets[0] == lines[1:4]
+
+    assert len(snippets[1]) == 4
+    assert snippets[1] == lines[3:7]
+
+
+def test_snippet_ids():
+    assert get_snippet_ids("") == []
+    assert get_snippet_ids("//! ") == []  # Invalid id
+    assert get_snippet_ids("//! [some name]") == ["some name"]
+    assert get_snippet_ids("//! [some name] [some other name]") == ["some name"]
+    assert get_snippet_ids("//! [some name] //! ") == ["some name"]  # Invalid id
+    assert get_snippet_ids("//! [some name] //! [some other name]") == ["some name", "some other name"]