Fix merging implicit multiline strings that have inline comments (#3956)

* Fix test behaviour

* Add new test cases

* Skip merging strings that have inline comments

* Don't merge lines with multiline strings with inline comments

* Changelog entry

* Document implicit multiline string merging rules

* Fix PR number

diff --git a/CHANGES.md b/CHANGES.md
index 2a50e45655042f24c2219fdab0bce5ef7eb6ba4f..79b5c6034e8ad2044f1417a5ccd1c0f4ee1cdf25 100644 (file)
--- a/CHANGES.md
+++ b/CHANGES.md
@@ -12,7 +12,7 @@
 
 ### Preview style
 
-<!-- Changes that affect Black's preview style -->
+- Fix merging implicit multiline strings that have inline comments (#3956)
 
 ### Configuration
 

diff --git a/docs/the_black_code_style/future_style.md b/docs/the_black_code_style/future_style.md
index 861bb64bff4d10016fde5db5230d887fbca364ec..367ff98537c8ff333bd919b62d0411e5f2d9732e 100644 (file)
--- a/docs/the_black_code_style/future_style.md
+++ b/docs/the_black_code_style/future_style.md
@@ -160,3 +160,67 @@ MULTILINE = """
 foobar
 """.replace("\n", "")
 ```
+
+Implicit multiline strings are special, because they can have inline comments. Strings
+without comments are merged, for example
+
+```python
+s = (
+    "An "
+    "implicit "
+    "multiline "
+    "string"
+)
+```
+
+becomes
+
+```python
+s = "An implicit multiline string"
+```
+
+A comment on any line of the string (or between two string lines) will block the
+merging, so
+
+```python
+s = (
+    "An "  # Important comment concerning just this line
+    "implicit "
+    "multiline "
+    "string"
+)
+```
+
+and
+
+```python
+s = (
+    "An "
+    "implicit "
+    # Comment in between
+    "multiline "
+    "string"
+)
+```
+
+will not be merged. Having the comment after or before the string lines (but still
+inside the parens) will merge the string. For example
+
+```python
+s = (  # Top comment
+    "An "
+    "implicit "
+    "multiline "
+    "string"
+    # Bottom comment
+)
+```
+
+becomes
+
+```python
+s = (  # Top comment
+    "An implicit multiline string"
+    # Bottom comment
+)
+```

diff --git a/src/black/linegen.py b/src/black/linegen.py
index d12ca39d0372972cef62043b230fa5785d67bd7f..2bfe587fa0ec5120287d133c7371ee85716d5ec7 100644 (file)
--- a/src/black/linegen.py
+++ b/src/black/linegen.py
@@ -587,6 +587,7 @@ def transform_line(
             or line.contains_unsplittable_type_ignore()
         )
         and not (line.inside_brackets and line.contains_standalone_comments())
+        and not line.contains_implicit_multiline_string_with_comments()
     ):
         # Only apply basic string preprocessing, since lines shouldn't be split here.
         if Preview.string_processing in mode:

diff --git a/src/black/lines.py b/src/black/lines.py
index 48fde88820853deec0e5b6405a8b4bf116bf9fe4..6acc95e7a7e004dbf2e35a571bb08def6f1100fe 100644 (file)
--- a/src/black/lines.py
+++ b/src/black/lines.py
@@ -239,6 +239,21 @@ class Line:
 
         return False
 
+    def contains_implicit_multiline_string_with_comments(self) -> bool:
+        """Chck if we have an implicit multiline string with comments on the line"""
+        for leaf_type, leaf_group_iterator in itertools.groupby(
+            self.leaves, lambda leaf: leaf.type
+        ):
+            if leaf_type != token.STRING:
+                continue
+            leaf_list = list(leaf_group_iterator)
+            if len(leaf_list) == 1:
+                continue
+            for leaf in leaf_list:
+                if self.comments_after(leaf):
+                    return True
+        return False
+
     def contains_uncollapsable_type_comments(self) -> bool:
         ignored_ids = set()
         try:

diff --git a/src/black/trans.py b/src/black/trans.py
index a2bff7f227adb51bf1afa96242393384041374ed..a3f6467cc9e3920758ce3b5ddf00a5e2c41129b6 100644 (file)
--- a/src/black/trans.py
+++ b/src/black/trans.py
@@ -390,7 +390,19 @@ class StringMerger(StringTransformer, CustomSplitMapMixin):
                 and is_valid_index(idx + 1)
                 and LL[idx + 1].type == token.STRING
             ):
-                if not is_part_of_annotation(leaf):
+                # Let's check if the string group contains an inline comment
+                # If we have a comment inline, we don't merge the strings
+                contains_comment = False
+                i = idx
+                while is_valid_index(i):
+                    if LL[i].type != token.STRING:
+                        break
+                    if line.comments_after(LL[i]):
+                        contains_comment = True
+                        break
+                    i += 1
+
+                if not is_part_of_annotation(leaf) and not contains_comment:
                     string_indices.append(idx)
 
                 # Advance to the next non-STRING leaf.

diff --git a/tests/data/cases/preview_long_strings__regression.py b/tests/data/cases/preview_long_strings__regression.py
index 40d5e745cc82bdaf091cfa71feded01c77724f41..436157f4e0584b949c109db62b3a0bad3fb97cf0 100644 (file)
--- a/tests/data/cases/preview_long_strings__regression.py
+++ b/tests/data/cases/preview_long_strings__regression.py
@@ -210,8 +210,8 @@ def foo():
 
 some_tuple = ("some string", "some string" " which should be joined")
 
-some_commented_string = (
-    "This string is long but not so long that it needs hahahah toooooo be so greatttt"  # This comment gets thrown to the top.
+some_commented_string = (  # This comment stays at the top.
+    "This string is long but not so long that it needs hahahah toooooo be so greatttt"
     " {} that I just can't think of any more good words to say about it at"
     " allllllllllll".format("ha")  # comments here are fine
 )
@@ -834,7 +834,7 @@ def foo():
 
 some_tuple = ("some string", "some string which should be joined")
 
-some_commented_string = (  # This comment gets thrown to the top.
+some_commented_string = (  # This comment stays at the top.
     "This string is long but not so long that it needs hahahah toooooo be so greatttt"
     " {} that I just can't think of any more good words to say about it at"
     " allllllllllll".format("ha")  # comments here are fine

diff --git a/tests/data/cases/preview_multiline_strings.py b/tests/data/cases/preview_multiline_strings.py
index dec4ef2e548d0095b275035a4e74d36abe36b0c2..3ff643610b7d50cd190f9b17056e8b28f985ae5f 100644 (file)
--- a/tests/data/cases/preview_multiline_strings.py
+++ b/tests/data/cases/preview_multiline_strings.py
@@ -157,6 +157,24 @@ Please use `--build-option` instead,
 `--global-option` is reserved to flags like `--verbose` or `--quiet`.
 """
 
+this_will_become_one_line = (
+    "a"
+    "b"
+    "c"
+)
+
+this_will_stay_on_three_lines = (
+    "a"  # comment
+    "b"
+    "c"
+)
+
+this_will_also_become_one_line = (  # comment
+    "a"
+    "b"
+    "c"
+)
+
 # output
 """cow
 say""",
@@ -357,3 +375,13 @@ msg = f"""The arguments {bad_arguments} were passed in.
 Please use `--build-option` instead,
 `--global-option` is reserved to flags like `--verbose` or `--quiet`.
 """
+
+this_will_become_one_line = "abc"
+
+this_will_stay_on_three_lines = (
+    "a"  # comment
+    "b"
+    "c"
+)
+
+this_will_also_become_one_line = "abc"  # comment