From d1b835799f85ba4fcdce2b2c980df03b48b056c5 Mon Sep 17 00:00:00 2001
From: David Roe <dave.roe@bearer.sh>
Date: Thu, 23 May 2024 12:21:14 +0100
Subject: [PATCH] fix(python): handle string line continuation (#1610)

---
 .../TestPythonString-string_literal           | 35 ++++++++++++++++++-
 .../python/detectors/string/string.go         | 14 +++++---
 .../detectors/testdata/string_literal.py      |  3 ++
 3 files changed, 47 insertions(+), 5 deletions(-)

diff --git a/internal/languages/python/detectors/.snapshots/TestPythonString-string_literal b/internal/languages/python/detectors/.snapshots/TestPythonString-string_literal
index c10474f24..6996b96ac 100644
--- a/internal/languages/python/detectors/.snapshots/TestPythonString-string_literal
+++ b/internal/languages/python/detectors/.snapshots/TestPythonString-string_literal
@@ -1,6 +1,6 @@
 type: module
 id: 0
-range: 1:1 - 12:1
+range: 1:1 - 14:3
 dataflow_sources:
     - 1
     - 5
@@ -12,6 +12,7 @@ dataflow_sources:
     - 37
     - 42
     - 49
+    - 58
 children:
     - type: expression_statement
       id: 1
@@ -284,6 +285,31 @@ children:
                 - type: '"""'
                   id: 57
                   range: 11:14 - 11:15
+    - type: expression_statement
+      id: 58
+      range: 13:1 - 14:3
+      dataflow_sources:
+        - 59
+      children:
+        - type: string
+          id: 59
+          range: 13:1 - 14:3
+          dataflow_sources:
+            - 60
+            - 61
+            - 62
+          children:
+            - type: '"""'
+              id: 60
+              range: 13:1 - 13:2
+            - type: escape_sequence
+              id: 61
+              range: 13:3 - 14:1
+              content: |
+                \
+            - type: '"""'
+              id: 62
+              range: 14:2 - 14:3
 
 - node: 2
   content: '''a'''
@@ -342,6 +368,13 @@ children:
   data:
     value: foo
     isliteral: true
+- node: 59
+  content: |-
+    "a\
+    b"
+  data:
+    value: ab
+    isliteral: true
 - node: 11
   content: '"a"'
   data:
diff --git a/internal/languages/python/detectors/string/string.go b/internal/languages/python/detectors/string/string.go
index 95b90425d..b3b65b896 100644
--- a/internal/languages/python/detectors/string/string.go
+++ b/internal/languages/python/detectors/string/string.go
@@ -77,12 +77,18 @@ func handleTemplateString(node *tree.Node, detectorContext types.Context) ([]int
 
 		switch {
 		case child.Type() == "escape_sequence":
-			value, err := stringutil.Unescape(child.Content())
-			if err != nil {
-				return fmt.Errorf("failed to decode escape sequence: %w", err)
+			// tree sitter parser doesn't handle line continuation inside a string
+			if child.Content() == "\\\n" || child.Content() == "\\\r\n" {
+				childValue = ""
+			} else {
+				value, err := stringutil.Unescape(child.Content())
+				if err != nil {
+					return fmt.Errorf("failed to decode escape sequence '%s': %w", child.Content(), err)
+				}
+
+				childValue = value
 			}
 
-			childValue = value
 			childIsLiteral = true
 		case len(namedChildren) == 0:
 			childValue = ""
diff --git a/internal/languages/python/detectors/testdata/string_literal.py b/internal/languages/python/detectors/testdata/string_literal.py
index 058723bc2..1a4cc4522 100644
--- a/internal/languages/python/detectors/testdata/string_literal.py
+++ b/internal/languages/python/detectors/testdata/string_literal.py
@@ -9,3 +9,6 @@
 
 False or "foo"
 "hey" or "foo"
+
+"a\
+b"
\ No newline at end of file