Skip to content

graflo.db.postgres.fuzzy_matcher

Fuzzy matching utilities for PostgreSQL schema analysis.

This module provides improved fuzzy matching strategies for identifying vertex names from table and column fragments.

FuzzyMatchCache

Cache for fuzzy matching fragments to vertex names.

Pre-computes fuzzy matches for all fragments to avoid redundant computations. This significantly improves performance when processing multiple tables.

Source code in graflo/db/postgres/fuzzy_matcher.py 
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
class FuzzyMatchCache:
    """Cache for fuzzy matching fragments to vertex names.

    Pre-computes fuzzy matches for all fragments to avoid redundant computations.
    This significantly improves performance when processing multiple tables.
    """

    def __init__(self, vertex_names: list[str], threshold: float = 0.6):
        """Initialize the fuzzy match cache.

        Args:
            vertex_names: List of vertex table names to match against
            threshold: Similarity threshold (0.0 to 1.0)
        """
        self.vertex_names = vertex_names
        self.threshold = threshold
        self._matcher = FuzzyMatcher(vertex_names, threshold)
        self._cache: dict[str, str | None] = {}
        self._build_cache()

    def _build_cache(self) -> None:
        """Pre-compute fuzzy matches for common patterns."""
        # Pre-compute exact matches (case-insensitive)
        for vertex_name in self.vertex_names:
            vertex_lower = vertex_name.lower()
            self._cache[vertex_lower] = vertex_name
            # Also cache common variations
            for suffix in ["id", "fk", "key", "pk", "ref", "reference"]:
                self._cache[f"{vertex_lower}_{suffix}"] = vertex_name
                self._cache[f"{suffix}_{vertex_lower}"] = vertex_name

    def get_match(self, fragment: str) -> str | None:
        """Get cached fuzzy match for a fragment, computing if not cached.

        Args:
            fragment: Fragment to match

        Returns:
            Best matching vertex name or None if no match above threshold
        """
        fragment_lower = fragment.lower()

        # Check cache first
        if fragment_lower in self._cache:
            return self._cache[fragment_lower]

        # Compute match if not cached using improved matcher
        match, _ = self._matcher.match(fragment)
        self._cache[fragment_lower] = match
        return match

    def batch_match(self, fragments: list[str]) -> dict[str, str | None]:
        """Match multiple fragments in batch, using cache when possible.

        Args:
            fragments: List of fragments to match

        Returns:
            Dictionary mapping fragments to their matched vertex names (or None)
        """
        results = {}
        for fragment in fragments:
            results[fragment] = self.get_match(fragment)
        return results

__init__(vertex_names, threshold=0.6)

Initialize the fuzzy match cache.

Parameters:

Name Type Description Default
vertex_names list[str]

List of vertex table names to match against

 required
threshold float

Similarity threshold (0.0 to 1.0)

 0.6
Source code in graflo/db/postgres/fuzzy_matcher.py 
225
226
227
228
229
230
231
232
233
234
235
236
def __init__(self, vertex_names: list[str], threshold: float = 0.6):
    """Initialize the fuzzy match cache.

    Args:
        vertex_names: List of vertex table names to match against
        threshold: Similarity threshold (0.0 to 1.0)
    """
    self.vertex_names = vertex_names
    self.threshold = threshold
    self._matcher = FuzzyMatcher(vertex_names, threshold)
    self._cache: dict[str, str | None] = {}
    self._build_cache()

batch_match(fragments)

Match multiple fragments in batch, using cache when possible.

Parameters:

Name Type Description Default
fragments list[str]

List of fragments to match

 required

Returns:

Type Description
dict[str, str | None]

Dictionary mapping fragments to their matched vertex names (or None)
Source code in graflo/db/postgres/fuzzy_matcher.py 
269
270
271
272
273
274
275
276
277
278
279
280
281
def batch_match(self, fragments: list[str]) -> dict[str, str | None]:
    """Match multiple fragments in batch, using cache when possible.

    Args:
        fragments: List of fragments to match

    Returns:
        Dictionary mapping fragments to their matched vertex names (or None)
    """
    results = {}
    for fragment in fragments:
        results[fragment] = self.get_match(fragment)
    return results

get_match(fragment)

Get cached fuzzy match for a fragment, computing if not cached.

Parameters:

Name Type Description Default
fragment str

Fragment to match

 required

Returns:

Type Description
str | None

Best matching vertex name or None if no match above threshold
Source code in graflo/db/postgres/fuzzy_matcher.py 
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
def get_match(self, fragment: str) -> str | None:
    """Get cached fuzzy match for a fragment, computing if not cached.

    Args:
        fragment: Fragment to match

    Returns:
        Best matching vertex name or None if no match above threshold
    """
    fragment_lower = fragment.lower()

    # Check cache first
    if fragment_lower in self._cache:
        return self._cache[fragment_lower]

    # Compute match if not cached using improved matcher
    match, _ = self._matcher.match(fragment)
    self._cache[fragment_lower] = match
    return match

FuzzyMatcher

Improved fuzzy matcher with multiple matching strategies.

Uses a combination of matching techniques: 1. Exact matching (case-insensitive) 2. Substring matching (with length-based scoring) 3. Sequence similarity (difflib) 4. Prefix/suffix matching 5. Common pattern matching (handles id, fk, etc.)

Source code in graflo/db/postgres/fuzzy_matcher.py 
 10
 11
 12
 13
 14
 15
 16
 17
 18
 19
 20
 21
 22
 23
 24
 25
 26
 27
 28
 29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
class FuzzyMatcher:
    """Improved fuzzy matcher with multiple matching strategies.

    Uses a combination of matching techniques:
    1. Exact matching (case-insensitive)
    2. Substring matching (with length-based scoring)
    3. Sequence similarity (difflib)
    4. Prefix/suffix matching
    5. Common pattern matching (handles id, fk, etc.)
    """

    def __init__(self, vertex_names: list[str], threshold: float = 0.6):
        """Initialize the fuzzy matcher.

        Args:
            vertex_names: List of vertex table names to match against
            threshold: Similarity threshold (0.0 to 1.0)
        """
        self.vertex_names = vertex_names
        self.threshold = threshold
        # Pre-compute lowercase versions for efficiency
        self._vertex_lower_map = {vn.lower(): vn for vn in vertex_names}
        self._vertex_lower_list = list(self._vertex_lower_map.keys())

    def match(self, fragment: str) -> tuple[str | None, float]:
        """Match a fragment against vertex names using multiple strategies.

        Args:
            fragment: Fragment to match

        Returns:
            Tuple of (best_match, score) or (None, 0.0) if no match above threshold
        """
        if not self.vertex_names or not fragment:
            return (None, 0.0)

        fragment_lower = fragment.lower()

        # Strategy 1: Exact match (highest priority, returns immediately)
        if fragment_lower in self._vertex_lower_map:
            return (self._vertex_lower_map[fragment_lower], 1.0)

        best_match = None
        best_score = 0.0

        # Strategy 2: Substring matching with length-based scoring
        substring_score = self._substring_match(fragment_lower)
        if substring_score[1] > best_score:
            best_match, best_score = substring_score

        # Strategy 3: Sequence similarity (difflib)
        sequence_score = self._sequence_match(fragment_lower)
        if sequence_score[1] > best_score:
            best_match, best_score = sequence_score

        # Strategy 4: Prefix/suffix matching
        prefix_suffix_score = self._prefix_suffix_match(fragment_lower)
        if prefix_suffix_score[1] > best_score:
            best_match, best_score = prefix_suffix_score

        # Strategy 5: Common pattern matching (handles id, fk, etc.)
        pattern_score = self._pattern_match(fragment_lower)
        if pattern_score[1] > best_score:
            best_match, best_score = pattern_score

        # Return match only if above threshold
        if best_score >= self.threshold:
            return (best_match, best_score)
        return (None, 0.0)

    def _substring_match(self, fragment_lower: str) -> tuple[str | None, float]:
        """Match using substring containment with length-based scoring.

        Args:
            fragment_lower: Lowercase fragment to match

        Returns:
            Tuple of (best_match, score)
        """
        best_match = None
        best_score = 0.0

        for vertex_lower, vertex_name in self._vertex_lower_map.items():
            # Check if fragment is contained in vertex or vice versa
            if fragment_lower in vertex_lower:
                # Fragment is substring of vertex (e.g., "user" in "users")
                score = len(fragment_lower) / len(vertex_lower)
                # Boost score if fragment is significant portion
                if len(fragment_lower) >= 3:  # At least 3 chars
                    score = min(score * 1.2, 0.95)  # Cap at 0.95
                if score > best_score:
                    best_score = score
                    best_match = vertex_name
            elif vertex_lower in fragment_lower:
                # Vertex is substring of fragment (e.g., "user" in "user_id")
                score = len(vertex_lower) / len(fragment_lower)
                # Boost score if vertex is significant portion
                if len(vertex_lower) >= 3:
                    score = min(score * 1.2, 0.95)
                if score > best_score:
                    best_score = score
                    best_match = vertex_name

        return (best_match, best_score)

    def _sequence_match(self, fragment_lower: str) -> tuple[str | None, float]:
        """Match using sequence similarity (difflib).

        Args:
            fragment_lower: Lowercase fragment to match

        Returns:
            Tuple of (best_match, score)
        """
        best_match = None
        best_score = 0.0

        for vertex_lower, vertex_name in self._vertex_lower_map.items():
            similarity = SequenceMatcher(None, fragment_lower, vertex_lower).ratio()
            if similarity > best_score:
                best_score = similarity
                best_match = vertex_name

        return (best_match, best_score)

    def _prefix_suffix_match(self, fragment_lower: str) -> tuple[str | None, float]:
        """Match using prefix or suffix patterns.

        Args:
            fragment_lower: Lowercase fragment to match

        Returns:
            Tuple of (best_match, score)
        """
        best_match = None
        best_score = 0.0

        for vertex_lower, vertex_name in self._vertex_lower_map.items():
            # Check prefix match
            if fragment_lower.startswith(vertex_lower):
                score = len(vertex_lower) / len(fragment_lower)
                if score > best_score:
                    best_score = score
                    best_match = vertex_name
            # Check suffix match
            elif fragment_lower.endswith(vertex_lower):
                score = len(vertex_lower) / len(fragment_lower)
                if score > best_score:
                    best_score = score
                    best_match = vertex_name
            # Check if vertex starts with fragment
            elif vertex_lower.startswith(fragment_lower):
                score = len(fragment_lower) / len(vertex_lower)
                if score > best_score:
                    best_score = score
                    best_match = vertex_name

        return (best_match, best_score)

    def _pattern_match(self, fragment_lower: str) -> tuple[str | None, float]:
        """Match using common patterns (id, fk, etc.).

        Args:
            fragment_lower: Lowercase fragment to match

        Returns:
            Tuple of (best_match, score)
        """
        # Common suffixes/prefixes to remove
        common_patterns = [
            ("_id", ""),
            ("_fk", ""),
            ("_key", ""),
            ("_pk", ""),
            ("_ref", ""),
            ("_reference", ""),
            ("id_", ""),
            ("fk_", ""),
            ("key_", ""),
            ("pk_", ""),
            ("ref_", ""),
            ("reference_", ""),
        ]

        best_match = None
        best_score = 0.0

        # Try removing common patterns and matching
        for pattern, replacement in common_patterns:
            if fragment_lower.endswith(pattern):
                base = fragment_lower[: -len(pattern)]
                if base in self._vertex_lower_map:
                    # High score for pattern-based matches
                    score = 0.9
                    if score > best_score:
                        best_score = score
                        best_match = self._vertex_lower_map[base]
            elif fragment_lower.startswith(pattern):
                base = fragment_lower[len(pattern) :]
                if base in self._vertex_lower_map:
                    score = 0.9
                    if score > best_score:
                        best_score = score
                        best_match = self._vertex_lower_map[base]

        return (best_match, best_score)

__init__(vertex_names, threshold=0.6)

Initialize the fuzzy matcher.

Parameters:

Name Type Description Default
vertex_names list[str]

List of vertex table names to match against

 required
threshold float

Similarity threshold (0.0 to 1.0)

 0.6
Source code in graflo/db/postgres/fuzzy_matcher.py 
21
22
23
24
25
26
27
28
29
30
31
32
def __init__(self, vertex_names: list[str], threshold: float = 0.6):
    """Initialize the fuzzy matcher.

    Args:
        vertex_names: List of vertex table names to match against
        threshold: Similarity threshold (0.0 to 1.0)
    """
    self.vertex_names = vertex_names
    self.threshold = threshold
    # Pre-compute lowercase versions for efficiency
    self._vertex_lower_map = {vn.lower(): vn for vn in vertex_names}
    self._vertex_lower_list = list(self._vertex_lower_map.keys())

match(fragment)

Match a fragment against vertex names using multiple strategies.

Parameters:

Name Type Description Default
fragment str

Fragment to match

 required

Returns:

Type Description
tuple[str | None, float]

Tuple of (best_match, score) or (None, 0.0) if no match above threshold
Source code in graflo/db/postgres/fuzzy_matcher.py 
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
def match(self, fragment: str) -> tuple[str | None, float]:
    """Match a fragment against vertex names using multiple strategies.

    Args:
        fragment: Fragment to match

    Returns:
        Tuple of (best_match, score) or (None, 0.0) if no match above threshold
    """
    if not self.vertex_names or not fragment:
        return (None, 0.0)

    fragment_lower = fragment.lower()

    # Strategy 1: Exact match (highest priority, returns immediately)
    if fragment_lower in self._vertex_lower_map:
        return (self._vertex_lower_map[fragment_lower], 1.0)

    best_match = None
    best_score = 0.0

    # Strategy 2: Substring matching with length-based scoring
    substring_score = self._substring_match(fragment_lower)
    if substring_score[1] > best_score:
        best_match, best_score = substring_score

    # Strategy 3: Sequence similarity (difflib)
    sequence_score = self._sequence_match(fragment_lower)
    if sequence_score[1] > best_score:
        best_match, best_score = sequence_score

    # Strategy 4: Prefix/suffix matching
    prefix_suffix_score = self._prefix_suffix_match(fragment_lower)
    if prefix_suffix_score[1] > best_score:
        best_match, best_score = prefix_suffix_score

    # Strategy 5: Common pattern matching (handles id, fk, etc.)
    pattern_score = self._pattern_match(fragment_lower)
    if pattern_score[1] > best_score:
        best_match, best_score = pattern_score

    # Return match only if above threshold
    if best_score >= self.threshold:
        return (best_match, best_score)
    return (None, 0.0)