Various changes.

[python_utils.git] / string_utils.py

diff --git a/string_utils.py b/string_utils.py
index 6fc257de52c48f34e207e79e8b2227e914ad2b8c..bca2b70d5cd18bc8bb1198782d356f2707c1cbd5 100644 (file)
--- a/string_utils.py
+++ b/string_utils.py
@@ -9,7 +9,7 @@ import logging
 import random
 import re
 import string
-from typing import Any, Callable, List, Optional
+from typing import Any, Callable, Dict, Iterable, List, Optional, Tuple
 import unicodedata
 from uuid import uuid4
 
@@ -963,3 +963,55 @@ def thify(n: int) -> str:
         return "rd"
     else:
         return "th"
+
+
+def ngrams(txt: str, n: int):
+    words = txt.split()
+    return ngrams_presplit(words, n)
+
+
+def ngrams_presplit(words: Iterable[str], n: int):
+    for ngram in zip(*[words[i:] for i in range(n)]):
+        yield(' '.join(ngram))
+
+
+def bigrams(txt: str):
+    return ngrams(txt, 2)
+
+
+def trigrams(txt: str):
+    return ngrams(txt, 3)
+
+
+def shuffle_columns(
+        txt: Iterable[str],
+        specs: Iterable[Iterable[int]],
+        delim=''
+) -> Iterable[str]:
+    out = []
+    for spec in specs:
+        chunk = ''
+        for n in spec:
+            chunk = chunk + delim + txt[n]
+        chunk = chunk.strip(delim)
+        out.append(chunk)
+    return out
+
+
+def shuffle_columns_into_dict(
+        txt: Iterable[str],
+        specs: Iterable[Tuple[str, Iterable[int]]],
+        delim=''
+) -> Dict[str, str]:
+    out = {}
+    for spec in specs:
+        chunk = ''
+        for n in spec[1]:
+            chunk = chunk + delim + txt[n]
+        chunk = chunk.strip(delim)
+        out[spec[0]] = chunk
+    return out
+
+
+def interpolate_using_dict(txt: str, values: Dict[str, str]) -> str:
+    return sprintf(txt.format(**values), end='')