Transformations: Support escaped characters in key-value pair parsing (#47901)

* Additional logic to handle quoted values

* Simple test with quoted values that include spaces

* Update test description

* Updating logic to account for nested quotes

- Adding additional test for nested quotes

* Strip out line breaks and carriage returns pre-processing

* Fix typo in test result

* Update key-value logic to avoid regexp

- Minor changes to account for null values

* Correct escaping on test

* Additional tests

- Test for null values
- Test for nested separator characters
- Update quoting

public/app/features/transformers/extractFields/fieldExtractor.test.ts

@@ -14,6 +14,66 @@ describe('Extract fields from text', () => {
     `);
   });
 
+  it('Test key-values with single/double quotes', async () => {
+    const extractor = fieldExtractors.get(FieldExtractorID.KeyValues);
+    const out = extractor.parse('a="1",   "b"=\'2\',c=3  x:y ;\r\nz="d and 4"');
+    expect(out).toMatchInlineSnapshot(`
+      Object {
+        "a": "1",
+        "b": "2",
+        "c": "3",
+        "x": "y",
+        "z": "d and 4",
+      }
+    `);
+  });
+
+  it('Test key-values with nested single/double quotes', async () => {
+    const extractor = fieldExtractors.get(FieldExtractorID.KeyValues);
+    const out = extractor.parse(
+      `a="1",   "b"=\'2\',c=3  x:y ;\r\nz="dbl_quotes=\\"Double Quotes\\" sgl_quotes='Single Quotes'"`
+    );
+
+    expect(out).toMatchInlineSnapshot(`
+      Object {
+        "a": "1",
+        "b": "2",
+        "c": "3",
+        "x": "y",
+        "z": "dbl_quotes=\\"Double Quotes\\" sgl_quotes='Single Quotes'",
+      }
+    `);
+  });
+
+  it('Test key-values with nested separator characters', async () => {
+    const extractor = fieldExtractors.get(FieldExtractorID.KeyValues);
+    const out = extractor.parse(`a="1",   "b"=\'2\',c=3  x:y ;\r\nz="This is; testing& validating, 1=:2"`);
+
+    expect(out).toMatchInlineSnapshot(`
+      Object {
+        "a": "1",
+        "b": "2",
+        "c": "3",
+        "x": "y",
+        "z": "This is; testing& validating, 1=:2",
+      }
+    `);
+  });
+
+  it('Test key-values where some values are null', async () => {
+    const extractor = fieldExtractors.get(FieldExtractorID.KeyValues);
+    const out = extractor.parse(`a=, "b"=\'2\',c=3  x: `);
+
+    expect(out).toMatchInlineSnapshot(`
+      Object {
+        "a": "",
+        "b": "2",
+        "c": "3",
+        "x": "",
+      }
+    `);
+  });
+
   it('Split key+values', async () => {
     const extractor = fieldExtractors.get(FieldExtractorID.KeyValues);
     const out = extractor.parse('a="1",   "b"=\'2\',c=3  x:y ;\r\nz="7"');

public/app/features/transformers/extractFields/fieldExtractors.ts

@@ -19,33 +19,99 @@ const extJSON: FieldExtractor = {
   },
 };
 
-// strips quotes and leading/trailing braces in prom labels
-const stripDecor = /['"]|^\{|\}$/g;
-// splits on whitespace and other label pair delimiters
-const splitLines = /[\s,;&]+/g;
-// splits kv pairs
-const splitPair = /[=:]/g;
+function parseKeyValuePairs(raw: string): Record<string, string> {
+  const buff: string[] = []; // array of characters
+  let esc = '';
+  let key = '';
+  const obj: Record<string, string> = {};
+  for (let i = 0; i < raw.length; i++) {
+    let c = raw[i];
+    if (c === esc) {
+      esc = '';
+      c = raw[++i];
+    }
 
-const extLabels: FieldExtractor = {
-  id: FieldExtractorID.KeyValues,
-  name: 'Key+value pairs',
-  description: 'Look for a=b, c: d values in the line',
-  parse: (v: string) => {
-    const obj: Record<string, any> = {};
+    const isEscaped = c === '\\';
+    if (isEscaped) {
+      c = raw[++i];
+    }
+
+    // When escaped just append
+    if (isEscaped || esc.length) {
+      buff.push(c);
+      continue;
+    }
 
-    v.trim()
-      .replace(stripDecor, '')
-      .split(splitLines)
-      .forEach((pair) => {
-        let [k, v] = pair.split(splitPair);
+    if (c === `"` || c === `'`) {
+      esc = c;
+    }
 
-        if (k != null) {
-          obj[k] = v;
+    switch (c) {
+      case ':':
+      case '=':
+        if (buff.length) {
+          if (key) {
+            obj[key] = '';
+          }
+          key = buff.join('');
+          buff.length = 0; // clear values
         }
-      });
+        break;
 
-    return obj;
-  },
+      // escape chars
+      case `"`:
+      case `'`:
+      // whitespace
+      case ` `:
+      case `\n`:
+      case `\t`:
+      case `\r`:
+      case `\n`:
+        if (buff.length && key === '') {
+          obj[buff.join('')] = '';
+          buff.length = 0;
+        }
+      // seperators
+      case ',':
+      case ';':
+      case '&':
+      case '{':
+      case '}':
+        if (buff.length) {
+          const val = buff.join('');
+          if (key.length) {
+            obj[key] = val;
+            key = '';
+          } else {
+            key = val;
+          }
+          buff.length = 0; // clear values
+        }
+        break;
+
+      // append our buffer
+      default:
+        buff.push(c);
+        if (i === raw.length - 1) {
+          if (key === '' && buff.length) {
+            obj[buff.join('')] = '';
+            buff.length = 0;
+          }
+        }
+    }
+  }
+
+  if (key.length) {
+    obj[key] = buff.join('');
+  }
+  return obj;
+}
+
+const extLabels: FieldExtractor = {
+  id: FieldExtractorID.KeyValues,
+  name: 'Key+value pairs',
+  description: 'Look for a=b, c: d values in the line',
+  parse: parseKeyValuePairs,
 };
 
 const fmts = [extJSON, extLabels];