initial support for parsing tables

#8

.github/workflows/main.yml

@@ -42,8 +42,8 @@ jobs:
 
     - uses: actions/upload-artifact@v2
       with:
-        name: .mypy-coverage_${{ matrix.platform }}_${{ matrix.python-version }}
-        path: .mypy-coverage/
+        name: .coverage.mypy_${{ matrix.platform }}_${{ matrix.python-version }}
+        path: .coverage.mypy/
 
 
   pypi:

orgparse/extra.py

@@ -0,0 +1,74 @@
+import re
+from typing import List, Sequence, Dict, Iterator, Iterable, Union, Optional
+
+
+RE_TABLE_SEPARATOR = re.compile(r'\s*\|(\-+\+)*\-+\|')
+STRIP_CELL_WHITESPACE = True
+
+
+Row = Sequence[str]
+
+class Table:
+    def __init__(self, lines: List[str]) -> None:
+        self._lines = lines
+
+    @property
+    def blocks(self) -> Iterator[Sequence[Row]]:
+        group: List[Row] = []
+        first = True
+        for r in self._pre_rows():
+            if r is None:
+                if not first or len(group) > 0:
+                    yield group
+                    first = False
+                group = []
+            else:
+                group.append(r)
+        if len(group) > 0:
+            yield group
+
+    def __iter__(self) -> Iterator[Row]:
+        return self.rows
+
+    @property
+    def rows(self) -> Iterator[Row]:
+        for r in self._pre_rows():
+            if r is not None:
+                yield r
+
+    def _pre_rows(self) -> Iterator[Optional[Row]]:
+        for l in self._lines:
+            if RE_TABLE_SEPARATOR.match(l):
+                yield None
+            else:
+                pr = l.strip().strip('|').split('|')
+                if STRIP_CELL_WHITESPACE:
+                    pr = [x.strip() for x in pr]
+                yield pr
+        # TODO use iparse helper?
+
+    @property
+    def as_dicts(self) -> 'AsDictHelper':
+        bl = list(self.blocks)
+        if len(bl) != 2:
+            raise RuntimeError('Need two-block table to non-ambiguously guess column names')
+        hrows = bl[0]
+        if len(hrows) != 1:
+            raise RuntimeError(f'Need single row heading to guess column names, got: {hrows}')
+        columns = hrows[0]
+        assert len(set(columns)) == len(columns), f'Duplicate column names: {columns}'
+        return AsDictHelper(
+            columns=columns,
+            rows=bl[1],
+        )
+
+
+
+class AsDictHelper:
+    def __init__(self, columns: Sequence[str], rows: Sequence[Row]) -> None:
+        self.columns = columns
+        self._rows = rows
+
+    def __iter__(self) -> Iterator[Dict[str, str]]:
+        for x in self._rows:
+            yield {k: v for k, v in zip(self.columns, x)}

orgparse/tests/test_rich.py

@@ -0,0 +1,63 @@
+'''
+Tests for rich formatting: tables etc.
+'''
+from .. import load, loads
+from ..extra import Table
+
+import pytest # type: ignore
+
+
+def test_table() -> None:
+    root = loads('''
+|       |           |     |
+|       | "heading" |     |
+|       |           |     |
+|-------+-----------+-----|
+| reiwf | fef       |     |
+|-------+-----------+-----|
+|-------+-----------+-----|
+| aba   | caba      | 123 |
+| yeah  |           |   X |
+
+    |------------------------+-------|
+    | when                   | count |
+    | datetime               | int   |
+    |------------------------+-------|
+    |                        | -1    |
+    | [2020-11-05 Thu 23:44] |       |
+    | [2020-11-06 Fri 01:00] | 1     |
+    |------------------------+-------|
+
+some irrelevant text
+
+| simple |
+|--------|
+| value1 |
+| value2 |
+    ''')
+
+    # FIXME need to parse properly
+    t1 = Table(root._lines[1:10])
+    t2 = Table(root._lines[11:19])
+    t3 = Table(root._lines[22:26])
+
+    assert ilen(t1.blocks) == 4
+    assert list(t1.blocks)[2] == []
+    assert ilen(t1.rows) == 6
+
+    with pytest.raises(RuntimeError):
+        list(t1.as_dicts) # not sure what should it be
+
+    assert ilen(t2.blocks) == 2
+    assert ilen(t2.rows) == 5
+    assert list(t2.rows)[3] == ['[2020-11-05 Thu 23:44]', '']
+
+
+    assert ilen(t3.blocks) == 2
+    assert list(t3.rows) == [['simple'], ['value1'], ['value2']]
+    assert t3.as_dicts.columns == ['simple']
+    assert list(t3.as_dicts) == [{'simple': 'value1'}, {'simple': 'value2'}]
+
+
+def ilen(x) -> int:
+    return len(list(x))

tox.ini

@@ -15,6 +15,6 @@ commands =
     pip install -e .[linting]
     python -m mypy orgparse                          \
                    # txt report is a bit more convenient to view on CI
-                   --txt-report  .mypy-coverage      \
-                   --html-report .mypy-coverage      \
+                   --txt-report  .coverage.mypy      \
+                   --html-report .coverage.mypy      \
                    {posargs}