pystylometry 1.0.0__py3-none-any.whl → 1.3.0__py3-none-any.whl

pystylometry/character/character_metrics.py

@@ -5,9 +5,9 @@ in writing style. Character-level metrics are fundamental for authorship
 attribution and can reveal distinctive patterns in punctuation usage,
 word construction, and formatting preferences.
 
-Related GitHub Issue:
+Related GitHub Issues:
     #12 - Character-Level Metrics
-    https://github.com/craigtrim/pystylometry/issues/12
+    #27 - Native chunked analysis with Distribution dataclass
 
 Features implemented:
     - Average word length (characters per word)
@@ -26,114 +26,80 @@ References:
         JASIST, 60(3), 538-556.
 """
 
-from .._types import CharacterMetricsResult
-
-
-def compute_character_metrics(text: str) -> CharacterMetricsResult:
+import math
+
+from .._types import CharacterMetricsResult, Distribution, chunk_text, make_distribution
+
+# Character sets
+_PUNCTUATION = {
+    ".",
+    ",",
+    "!",
+    "?",
+    ";",
+    ":",
+    "-",
+    "—",
+    "–",  # Basic punctuation
+    "'",
+    '"',
+    """, """,
+    "'",
+    "'",  # Quotes
+    "(",
+    ")",
+    "[",
+    "]",
+    "{",
+    "}",  # Brackets
+    "/",
+    "\\",
+    "|",  # Slashes
+    "…",  # Ellipsis
+    "*",
+    "&",
+    "@",
+    "#",
+    "$",
+    "%",
+    "^",
+    "~",
+    "`",  # Special symbols
+}
+_VOWELS = {"a", "e", "i", "o", "u"}
+_STANDARD_LETTERS = set("abcdefghijklmnopqrstuvwxyz")
+
+
+def _compute_character_metrics_single(text: str) -> dict:
+    """Compute character-level metrics for a single chunk of text.
+
+    Returns a dict with all computed values, or values containing nan for empty text.
     """
-    Compute character-level stylometric metrics.
-
-    This function analyzes text at the character level to extract features
-    related to word length, punctuation usage, letter distribution, and
-    other low-level patterns that can be distinctive for authorship
-    attribution and style analysis.
-
-    Related GitHub Issue:
-        #12 - Character-Level Metrics
-        https://github.com/craigtrim/pystylometry/issues/12
-
-    Character-level features are particularly valuable because:
-        1. They are language-independent (work across languages)
-        2. They capture subconscious writing patterns
-        3. They are resistant to topic variation
-        4. They complement higher-level metrics (words, syntax)
-
-    Metrics computed:
-        - Average word length: Mean characters per word
-        - Average sentence length (chars): Mean characters per sentence
-        - Punctuation density: Punctuation marks per 100 words
-        - Punctuation variety: Count of unique punctuation types used
-        - Letter frequency: Distribution of a-z (case-insensitive)
-        - Vowel-to-consonant ratio: Ratio of vowels to consonants
-        - Digit count/ratio: Numeric character usage
-        - Uppercase ratio: Uppercase letters / total letters
-        - Whitespace ratio: Whitespace characters / total characters
-
-    Args:
-        text: Input text to analyze. Should contain at least one sentence
-              for meaningful results. Empty text will return NaN for ratios
-              and 0 for counts.
-
-    Returns:
-        CharacterMetricsResult with all character-level features and metadata.
-        For empty text, all ratios will be NaN and counts will be 0.
-
-    Example:
-        >>> result = compute_character_metrics("The quick brown fox jumps!")
-        >>> print(f"Avg word length: {result.avg_word_length:.2f}")
-        Avg word length: 4.17
-        >>> print(f"Punctuation density: {result.punctuation_density:.2f}")
-        Punctuation density: 16.67
-        >>> print(f"Vowel/consonant ratio: {result.vowel_consonant_ratio:.2f}")
-        Vowel/consonant ratio: 0.71
-
-        >>> # Empty text handling
-        >>> result = compute_character_metrics("")
-        >>> import math
-        >>> math.isnan(result.avg_word_length)
-        True
-        >>> result.digit_count
-        0
-
-    Note:
-        - Punctuation marks include: . , ! ? ; : - ' " ( ) [ ] { } ... etc.
-        - Whitespace includes spaces, tabs, newlines
-        - Letter frequency is case-insensitive (lowercase normalized)
-        - Words are tokenized by whitespace for length calculation
-        - Sentences are split using standard sentence delimiters (. ! ?)
-    """
-    # Define character sets
-    # GitHub Issue #12: https://github.com/craigtrim/pystylometry/issues/12
-    PUNCTUATION = {
-        ".", ",", "!", "?", ";", ":", "-", "—", "–",  # Basic punctuation
-        "'", '"', """, """, "'", "'",  # Quotes
-        "(", ")", "[", "]", "{", "}",  # Brackets
-        "/", "\\", "|",  # Slashes
-        "…",  # Ellipsis
-        "*", "&", "@", "#", "$", "%", "^", "~", "`",  # Special symbols
-    }
-    VOWELS = {"a", "e", "i", "o", "u"}
-
-    # Handle empty text
     if not text:
-        # Return NaN for all ratios, 0 for all counts
-        empty_letter_freq = {letter: 0.0 for letter in "abcdefghijklmnopqrstuvwxyz"}
-        return CharacterMetricsResult(
-            avg_word_length=float("nan"),
-            avg_sentence_length_chars=float("nan"),
-            punctuation_density=float("nan"),
-            punctuation_variety=0,
-            letter_frequency=empty_letter_freq,
-            vowel_consonant_ratio=float("nan"),
-            digit_count=0,
-            digit_ratio=float("nan"),
-            uppercase_ratio=float("nan"),
-            whitespace_ratio=float("nan"),
-            metadata={
-                "total_characters": 0,
-                "total_letters": 0,
-                "total_words": 0,
-                "total_sentences": 0,
-                "total_punctuation": 0,
-                "total_whitespace": 0,
-                "total_digits": 0,
-                "punctuation_types": [],
-                "vowel_count": 0,
-                "consonant_count": 0,
-                "uppercase_count": 0,
-                "lowercase_count": 0,
-            },
-        )
+        return {
+            "avg_word_length": float("nan"),
+            "avg_sentence_length_chars": float("nan"),
+            "punctuation_density": float("nan"),
+            "punctuation_variety": 0,
+            "vowel_consonant_ratio": float("nan"),
+            "digit_count": 0,
+            "digit_ratio": float("nan"),
+            "uppercase_ratio": float("nan"),
+            "whitespace_ratio": float("nan"),
+            "letter_frequency": {letter: 0.0 for letter in "abcdefghijklmnopqrstuvwxyz"},
+            "total_characters": 0,
+            "total_letters": 0,
+            "total_words": 0,
+            "total_sentences": 0,
+            "total_punctuation": 0,
+            "total_whitespace": 0,
+            "total_digits": 0,
+            "punctuation_types": [],
+            "vowel_count": 0,
+            "consonant_count": 0,
+            "uppercase_count": 0,
+            "lowercase_count": 0,
+        }
 
     # Initialize counters
     total_chars = len(text)
@@ -147,19 +113,18 @@ def compute_character_metrics(text: str) -> CharacterMetricsResult:
     punctuation_count = 0
     punctuation_types = set()
 
-    # Single pass through text to classify and count all characters
+    # Single pass through text
     for char in text:
         if char.isalpha():
-            # Letter - update letter frequency (case-insensitive)
-            letter_counts[char.lower()] += 1
+            lower_char = char.lower()
+            if lower_char in _STANDARD_LETTERS:
+                letter_counts[lower_char] += 1
 
-            # Count vowels and consonants
-            if char.lower() in VOWELS:
+            if lower_char in _VOWELS:
                 vowel_count += 1
-            else:
+            elif lower_char in _STANDARD_LETTERS:
                 consonant_count += 1
 
-            # Count uppercase and lowercase
             if char.isupper():
                 uppercase_count += 1
             else:
@@ -167,44 +132,35 @@ def compute_character_metrics(text: str) -> CharacterMetricsResult:
 
         elif char.isdigit():
             digit_count += 1
-
         elif char.isspace():
             whitespace_count += 1
-
-        elif char in PUNCTUATION:
+        elif char in _PUNCTUATION:
             punctuation_count += 1
             punctuation_types.add(char)
 
     total_letters = vowel_count + consonant_count
 
-    # Calculate letter frequency distribution (normalize to sum to 1.0)
+    # Letter frequency distribution
     if total_letters > 0:
-        letter_frequency = {letter: count / total_letters for letter, count in letter_counts.items()}
+        letter_frequency = {
+            letter: count / total_letters for letter, count in letter_counts.items()
+        }
     else:
         letter_frequency = {letter: 0.0 for letter in "abcdefghijklmnopqrstuvwxyz"}
 
-    # Tokenize into words (split on whitespace, then strip punctuation for length)
+    # Word metrics
     words = text.split()
     total_words = len(words)
 
-    # Calculate average word length (count only letters and digits in words)
     if total_words > 0:
-        word_lengths = []
-        for word in words:
-            # Count only alphanumeric characters for word length
-            word_length = sum(1 for char in word if char.isalnum())
-            if word_length > 0:  # Only count words with at least one alphanumeric char
-                word_lengths.append(word_length)
-
-        if word_lengths:
-            avg_word_length = sum(word_lengths) / len(word_lengths)
-        else:
-            avg_word_length = float("nan")
+        word_lengths = [
+            sum(1 for c in w if c.isalnum()) for w in words if any(c.isalnum() for c in w)
+        ]
+        avg_word_length = sum(word_lengths) / len(word_lengths) if word_lengths else float("nan")
     else:
         avg_word_length = float("nan")
 
-    # Segment text into sentences (split on . ! ?)
-    # Simple approach: split on sentence delimiters
+    # Sentence metrics
     sentence_delimiters = {".", "!", "?"}
     sentences = []
     current_sentence = []
@@ -212,13 +168,11 @@ def compute_character_metrics(text: str) -> CharacterMetricsResult:
     for char in text:
         current_sentence.append(char)
         if char in sentence_delimiters:
-            # End of sentence
             sentence_text = "".join(current_sentence).strip()
-            if sentence_text:  # Only add non-empty sentences
+            if sentence_text:
                 sentences.append(sentence_text)
             current_sentence = []
 
-    # Add any remaining text as a sentence if it's non-empty and doesn't end with delimiter
     if current_sentence:
         sentence_text = "".join(current_sentence).strip()
         if sentence_text:
@@ -226,52 +180,40 @@ def compute_character_metrics(text: str) -> CharacterMetricsResult:
 
     total_sentences = len(sentences)
 
-    # Calculate average sentence length in characters
     if total_sentences > 0:
         sentence_lengths = [len(sent) for sent in sentences]
         avg_sentence_length_chars = sum(sentence_lengths) / total_sentences
     else:
         avg_sentence_length_chars = float("nan")
 
-    # Calculate punctuation density (per 100 words)
-    if total_words > 0:
-        punctuation_density = (punctuation_count / total_words) * 100
-    else:
-        punctuation_density = float("nan")
-
-    # Punctuation variety (count of unique punctuation types)
+    # Ratios
+    punctuation_density = (
+        (punctuation_count / total_words * 100) if total_words > 0 else float("nan")
+    )
     punctuation_variety = len(punctuation_types)
 
-    # Calculate vowel-to-consonant ratio
     if consonant_count > 0:
         vowel_consonant_ratio = vowel_count / consonant_count
     elif vowel_count > 0:
-        # Vowels but no consonants - ratio is infinity
         vowel_consonant_ratio = float("inf")
     else:
-        # No letters at all
         vowel_consonant_ratio = float("nan")
 
-    # Calculate digit ratio
-    if total_chars > 0:
-        digit_ratio = digit_count / total_chars
-    else:
-        digit_ratio = float("nan")
-
-    # Calculate uppercase ratio
-    if total_letters > 0:
-        uppercase_ratio = uppercase_count / total_letters
-    else:
-        uppercase_ratio = float("nan")
-
-    # Calculate whitespace ratio
-    if total_chars > 0:
-        whitespace_ratio = whitespace_count / total_chars
-    else:
-        whitespace_ratio = float("nan")
-
-    # Build metadata
-    metadata = {
+    digit_ratio = digit_count / total_chars if total_chars > 0 else float("nan")
+    uppercase_ratio = uppercase_count / total_letters if total_letters > 0 else float("nan")
+    whitespace_ratio = whitespace_count / total_chars if total_chars > 0 else float("nan")
+
+    return {
+        "avg_word_length": avg_word_length,
+        "avg_sentence_length_chars": avg_sentence_length_chars,
+        "punctuation_density": punctuation_density,
+        "punctuation_variety": punctuation_variety,
+        "vowel_consonant_ratio": vowel_consonant_ratio,
+        "digit_count": digit_count,
+        "digit_ratio": digit_ratio,
+        "uppercase_ratio": uppercase_ratio,
+        "whitespace_ratio": whitespace_ratio,
+        "letter_frequency": letter_frequency,
         "total_characters": total_chars,
         "total_letters": total_letters,
         "total_words": total_words,
@@ -286,16 +228,162 @@ def compute_character_metrics(text: str) -> CharacterMetricsResult:
         "lowercase_count": lowercase_count,
     }
 
+
+def compute_character_metrics(text: str, chunk_size: int = 1000) -> CharacterMetricsResult:
+    """
+    Compute character-level stylometric metrics.
+
+    This function uses native chunked analysis to capture variance and patterns
+    across the text, which is essential for stylometric fingerprinting.
+
+    Related GitHub Issues:
+        #12 - Character-Level Metrics
+        #27 - Native chunked analysis with Distribution dataclass
+
+    Character-level features are particularly valuable because:
+        1. They are language-independent (work across languages)
+        2. They capture subconscious writing patterns
+        3. They are resistant to topic variation
+        4. They complement higher-level metrics (words, syntax)
+
+    Metrics computed:
+        - Average word length: Mean characters per word
+        - Average sentence length (chars): Mean characters per sentence
+        - Punctuation density: Punctuation marks per 100 words
+        - Punctuation variety: Count of unique punctuation types used
+        - Letter frequency: Distribution of a-z (case-insensitive)
+        - Vowel-to-consonant ratio: Ratio of vowels to consonants
+        - Digit count/ratio: Numeric character usage
+        - Uppercase ratio: Uppercase letters / total letters
+        - Whitespace ratio: Whitespace characters / total characters
+
+    Args:
+        text: Input text to analyze
+        chunk_size: Number of words per chunk (default: 1000)
+
+    Returns:
+        CharacterMetricsResult with all character-level features, distributions,
+        and metadata.
+
+    Example:
+        >>> result = compute_character_metrics("Long text...", chunk_size=1000)
+        >>> result.avg_word_length  # Mean across chunks
+        4.5
+        >>> result.avg_word_length_dist.std  # Variance reveals fingerprint
+        0.3
+    """
+    # Chunk the text
+    chunks = chunk_text(text, chunk_size)
+
+    # Compute metrics per chunk
+    chunk_results = [_compute_character_metrics_single(chunk) for chunk in chunks]
+
+    # Collect values for distributions
+    avg_word_length_vals = [
+        r["avg_word_length"] for r in chunk_results if not math.isnan(r["avg_word_length"])
+    ]
+    avg_sentence_vals = [
+        r["avg_sentence_length_chars"]
+        for r in chunk_results
+        if not math.isnan(r["avg_sentence_length_chars"])
+    ]
+    punct_density_vals = [
+        r["punctuation_density"] for r in chunk_results if not math.isnan(r["punctuation_density"])
+    ]
+    punct_variety_vals = [float(r["punctuation_variety"]) for r in chunk_results]
+    vc_ratio_vals = [
+        r["vowel_consonant_ratio"]
+        for r in chunk_results
+        if not math.isnan(r["vowel_consonant_ratio"]) and not math.isinf(r["vowel_consonant_ratio"])
+    ]
+    digit_ratio_vals = [r["digit_ratio"] for r in chunk_results if not math.isnan(r["digit_ratio"])]
+    uppercase_ratio_vals = [
+        r["uppercase_ratio"] for r in chunk_results if not math.isnan(r["uppercase_ratio"])
+    ]
+    whitespace_ratio_vals = [
+        r["whitespace_ratio"] for r in chunk_results if not math.isnan(r["whitespace_ratio"])
+    ]
+
+    # Aggregate totals
+    total_digits = sum(r["digit_count"] for r in chunk_results)
+    total_characters = sum(r["total_characters"] for r in chunk_results)
+    total_letters = sum(r["total_letters"] for r in chunk_results)
+    total_words = sum(r["total_words"] for r in chunk_results)
+    total_sentences = sum(r["total_sentences"] for r in chunk_results)
+    total_punctuation = sum(r["total_punctuation"] for r in chunk_results)
+    total_whitespace = sum(r["total_whitespace"] for r in chunk_results)
+    total_vowel_count = sum(r["vowel_count"] for r in chunk_results)
+    total_consonant_count = sum(r["consonant_count"] for r in chunk_results)
+    total_uppercase_count = sum(r["uppercase_count"] for r in chunk_results)
+    total_lowercase_count = sum(r["lowercase_count"] for r in chunk_results)
+    all_punctuation_types = set()
+    for r in chunk_results:
+        all_punctuation_types.update(r["punctuation_types"])
+
+    # Aggregate letter frequency
+    total_letter_counts = {letter: 0 for letter in "abcdefghijklmnopqrstuvwxyz"}
+    for r in chunk_results:
+        if r["total_letters"] > 0:
+            for letter, freq in r["letter_frequency"].items():
+                total_letter_counts[letter] += freq * r["total_letters"]
+
+    if total_letters > 0:
+        letter_frequency = {
+            letter: count / total_letters for letter, count in total_letter_counts.items()
+        }
+    else:
+        letter_frequency = {letter: 0.0 for letter in "abcdefghijklmnopqrstuvwxyz"}
+
+    # Build distributions (handle empty case)
+    def safe_dist(values: list[float]) -> Distribution:
+        if not values:
+            return Distribution(
+                values=[], mean=float("nan"), median=float("nan"), std=0.0, range=0.0, iqr=0.0
+            )
+        return make_distribution(values)
+
+    avg_word_length_dist = safe_dist(avg_word_length_vals)
+    avg_sentence_dist = safe_dist(avg_sentence_vals)
+    punct_density_dist = safe_dist(punct_density_vals)
+    punct_variety_dist = safe_dist(punct_variety_vals)
+    vc_ratio_dist = safe_dist(vc_ratio_vals)
+    digit_ratio_dist = safe_dist(digit_ratio_vals)
+    uppercase_ratio_dist = safe_dist(uppercase_ratio_vals)
+    whitespace_ratio_dist = safe_dist(whitespace_ratio_vals)
+
     return CharacterMetricsResult(
-        avg_word_length=avg_word_length,
-        avg_sentence_length_chars=avg_sentence_length_chars,
-        punctuation_density=punctuation_density,
-        punctuation_variety=punctuation_variety,
+        avg_word_length=avg_word_length_dist.mean,
+        avg_sentence_length_chars=avg_sentence_dist.mean,
+        punctuation_density=punct_density_dist.mean,
+        punctuation_variety=punct_variety_dist.mean,
         letter_frequency=letter_frequency,
-        vowel_consonant_ratio=vowel_consonant_ratio,
-        digit_count=digit_count,
-        digit_ratio=digit_ratio,
-        uppercase_ratio=uppercase_ratio,
-        whitespace_ratio=whitespace_ratio,
-        metadata=metadata,
+        vowel_consonant_ratio=vc_ratio_dist.mean,
+        digit_count=total_digits,
+        digit_ratio=digit_ratio_dist.mean,
+        uppercase_ratio=uppercase_ratio_dist.mean,
+        whitespace_ratio=whitespace_ratio_dist.mean,
+        avg_word_length_dist=avg_word_length_dist,
+        avg_sentence_length_chars_dist=avg_sentence_dist,
+        punctuation_density_dist=punct_density_dist,
+        punctuation_variety_dist=punct_variety_dist,
+        vowel_consonant_ratio_dist=vc_ratio_dist,
+        digit_ratio_dist=digit_ratio_dist,
+        uppercase_ratio_dist=uppercase_ratio_dist,
+        whitespace_ratio_dist=whitespace_ratio_dist,
+        chunk_size=chunk_size,
+        chunk_count=len(chunks),
+        metadata={
+            "total_characters": total_characters,
+            "total_letters": total_letters,
+            "total_words": total_words,
+            "total_sentences": total_sentences,
+            "total_punctuation": total_punctuation,
+            "total_whitespace": total_whitespace,
+            "total_digits": total_digits,
+            "punctuation_types": sorted(list(all_punctuation_types)),
+            "vowel_count": total_vowel_count,
+            "consonant_count": total_consonant_count,
+            "uppercase_count": total_uppercase_count,
+            "lowercase_count": total_lowercase_count,
+        },
     )