Remove tabulate dependency and optimize table formatting

snap/deciphon_snap/match.py

@@ -20,7 +20,7 @@ class MatchElemName(Enum):
     AMINO = 4
 
 
-@dataclass(slots=True, frozen=True)
+@dataclass(slots=True, frozen=True, match_args=False)
 class Match:
     raw: str
     start: int

snap/deciphon_snap/tabulate.py

@@ -0,0 +1,58 @@
+from itertools import zip_longest
+
+__all__ = ["tabulate"]
+
+
+def tabulate(tabular_data, aligns):
+    cols = list(zip_longest(*tabular_data))
+    cols = [["" if v is None else str(v) for v in c] for c in cols]
+    cols = [_align_column(c, a) for c, a in zip(cols, aligns)]
+    rows = list(zip(*cols))
+    return _format_table(rows)
+
+
+def _padleft(width, s):
+    fmt = "{0:>%ds}" % width
+    return fmt.format(s)
+
+
+def _padright(width, s):
+    fmt = "{0:<%ds}" % width
+    return fmt.format(s)
+
+
+def _align_column_choose_padfn(strings, alignment):
+    strings = [s.strip() for s in strings]
+    padfn = _padleft if alignment == "right" else _padright
+    return strings, padfn
+
+
+def _align_column(strings, alignment):
+    """[string] -> [padded_string]"""
+    strings, padfn = _align_column_choose_padfn(strings, alignment)
+
+    s_widths = list(map(len, strings))
+    maxwidth = max(s_widths)
+    return [padfn(maxwidth, s) for s in strings]
+
+
+def _build_row(padded_cells):
+    "Return a string which represents a row of data cells."
+    return " ".join(padded_cells).rstrip()
+
+
+def _append_basic_row(lines, padded_cells):
+    lines.append(_build_row(padded_cells))
+    return lines
+
+
+def _format_table(rows):
+    """Produce a plain-text representation of the table."""
+    lines = []
+
+    padded_rows = [[cell for cell in row] for row in rows]
+
+    for row in padded_rows:
+        _append_basic_row(lines, row)
+
+    return "\n".join(lines)

snap/deciphon_snap/view.py

@@ -5,11 +5,11 @@
 
 from hmmer_tables.query import DomAnnot, read_query
 from pydantic import BaseModel
-from tabulate import simple_separated_format, tabulate
 
 from deciphon_snap.match import Match, MatchList
 from deciphon_snap.prod import H3Result, Prod
 from deciphon_snap.snap_file import SnapFile
+from deciphon_snap.tabulate import tabulate
 
 __all__ = ["view_alignments"]
 
@@ -197,16 +197,10 @@ def grab_query(x, i):
             [None, None, "".join(score[sl]) + pad, "PP"],
         ]
         table += row + [[None, None, None, None]]
-    tablefmt = simple_separated_format(" ")
-    txt += tabulate(
-        table, tablefmt=tablefmt, colalign=("right", "right", "left", "left")
-    )
+    txt += tabulate(table, ["right", "right", "left", "left"])
     txt = txt.replace("&", "") + "\n"
     return txt
 
 
 def view_alignments(snap: SnapFile):
-    txt = []
-    for prod in snap.products:
-        txt.append(view_alignment(prod))
-    return "\n".join(txt)
+    return (view_alignment(prod) for prod in snap.products)

snap/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "deciphon-snap"
-version = "0.11.3"
+version = "0.12.0"
 description = "Reader for Deciphon snap files."
 authors = ["Danilo Horta <[email protected]>"]
 license = "MIT"
@@ -11,7 +11,6 @@ packages = [{ include = "deciphon_snap" }]
 python = "^3.9"
 fsspec = ">=2024.5.0"
 h3result = "^0.3"
-tabulate = "^0.9"
 fasta-reader = "^3.0"
 deciphon-intervals = "^0.1"
 prettytable = "^3.10"

snap/tests/test_alignments.py

@@ -3,7 +3,7 @@
 from deciphon_snap.read_snap import read_snap
 from deciphon_snap.view import view_alignments
 
+
 def test_alignments(files_path: Path):
     snap_file = read_snap(files_path / "consensus.dcs")
-    txt = view_alignments(snap_file)
-    print(txt)
+    print("\n".join(view_alignments(snap_file)))