mergeron 2025.739290.3__py3-none-any.whl → 2025.739290.4__py3-none-any.whl

mergeron/core/ftc_merger_investigations_data.py

@@ -10,25 +10,37 @@ Reported row and column totals from source data are not stored.
 
 from __future__ import annotations
 
+import re
 import shutil
-from collections.abc import Mapping, Sequence
-from dataclasses import dataclass
+from collections.abc import Sequence
 from importlib import resources
 from operator import itemgetter
 from pathlib import Path
 from types import MappingProxyType
 from typing import Any
+from zipfile import ZIP_DEFLATED, ZipFile
 
-import msgpack  # type: ignore
 import msgpack_numpy as m  # type: ignore
 import numpy as np
-import re2 as re  # type: ignore
 import urllib3
 from bs4 import BeautifulSoup
 from numpy.testing import assert_array_equal
-from ruamel import yaml
 
-from .. import _PKG_NAME, DATA_DIR, EMPTY_ARRAYINT, VERSION, ArrayBIGINT  # noqa: TID252
+from .. import (  # noqa: TID252
+    _PKG_NAME,
+    DATA_DIR,
+    EMPTY_ARRAYINT,
+    VERSION,
+    ArrayBIGINT,
+    this_yaml,
+)
+from . import (
+    INVData,
+    INVData_in_,
+    INVTableData,
+    _dict_from_mapping,
+    _mappingproxy_from_mapping,
+)
 
 __version__ = VERSION
 
@@ -38,7 +50,7 @@ FTCDATA_DIR = DATA_DIR / "FTCData"
 if not FTCDATA_DIR.is_dir():
     FTCDATA_DIR.mkdir(parents=True)
 
-INVDATA_ARCHIVE_PATH = DATA_DIR / "ftc_invdata.msgpack"
+INVDATA_ARCHIVE_PATH = DATA_DIR / "ftc_invdata.zip"
 if (
     not INVDATA_ARCHIVE_PATH.is_file()
     and (
@@ -93,32 +105,6 @@ CNT_FCOUNT_DICT = {
 }
 
 
-@dataclass(slots=True, frozen=True)
-class INVTableData:
-    industry_group: str
-    additional_evidence: str
-    data_array: ArrayBIGINT
-
-    @classmethod
-    def to_yaml(
-        cls, _r: yaml.representer.SafeRepresenter, _d: INVTableData
-    ) -> yaml.MappingNode:
-        _ret: yaml.MappingNode = _r.represent_mapping(
-            f"!{cls.__name__}", {_a: getattr(_d, _a) for _a in _d.__dataclass_fields__}
-        )
-        return _ret
-
-    @classmethod
-    def from_yaml(
-        cls, _c: yaml.constructor.SafeConstructor, _n: yaml.MappingNode
-    ) -> INVTableData:
-        return cls(**_c.construct_mapping(_n))
-
-
-type INVData = Mapping[str, Mapping[str, Mapping[str, INVTableData]]]
-type _INVData_in = dict[str, dict[str, dict[str, INVTableData]]]
-
-
 def construct_data(
     _archive_path: Path = INVDATA_ARCHIVE_PATH,
     *,
@@ -157,75 +143,83 @@ def construct_data(
     """
 
     if _archive_path.is_file() and not rebuild_data:
-        _archived_data = msgpack.unpackb(_archive_path.read_bytes(), use_list=False)
-
-        _invdata: _INVData_in = {}
-        for _period in _archived_data:
-            _invdata[_period] = {}
-            for _table_type in _archived_data[_period]:
-                _invdata[_period][_table_type] = {}
-                for _table_no in _archived_data[_period][_table_type]:
-                    _invdata[_period][_table_type][_table_no] = INVTableData(
-                        *_archived_data[_period][_table_type][_table_no]
-                    )
-        return MappingProxyType(_invdata)
+        with (
+            ZipFile(_archive_path, "r") as _yzh,
+            _yzh.open(f"{_archive_path.stem}.yaml", "r") as _yfh,
+        ):
+            invdata_ = this_yaml.load(_yfh)
+        if isinstance(invdata_, MappingProxyType):
+            invdata_ = _mappingproxy_from_mapping(invdata_)
+            with (
+                ZipFile(_archive_path, "w", compression=ZIP_DEFLATED) as _yzh,
+                _yzh.open(f"{_archive_path.stem}.yaml", "w") as _yfh,
+            ):
+                this_yaml.dump(invdata_, _yfh)
+        return invdata_
 
-    _invdata = dict(_parse_invdata())  # type: ignore # Convert immutable to mutable
+    invdata: INVData_in_ = _dict_from_mapping(_parse_invdata())
 
     # Add some data periods (
     #   only periods ending in 2011, others have few observations and
     #   some incompatibilities
     #   )
-    for _data_period in "2004-2011", "2006-2011", "2008-2011":
-        _invdata_bld = _construct_new_period_data(
-            _invdata,
-            _data_period,
+    for data_period in "2004-2011", "2006-2011", "2008-2011":
+        invdata_bld = _construct_new_period_data(
+            invdata,
+            data_period,
             flag_backward_compatibility=flag_backward_compatibility,
         )
-        _invdata |= {_data_period: _invdata_bld}
+        invdata |= {data_period: invdata_bld}
 
     # Create data for industries with no evidence on entry
-    for _data_period in _invdata:
-        _construct_no_evidence_data(_invdata, _data_period)
+    for data_period in invdata:
+        _construct_no_evidence_data(invdata, data_period)
 
     # Create a list of exclusions to named industries in the base period,
     #   for construction of aggregate enforcement statistics where feasible
-    _industry_exclusion_list = (
+    industry_exclusion_list = {
         "AllMarkets",
         "OtherMarkets",
         "IndustriesinCommon",
         "",
         ("PharmaceuticalsMarkets" if flag_pharma_for_exclusion else None),
-    )
-    for _data_period in "1996-2003", "1996-2011", "2004-2011":
-        for _table_type, _table_no in zip(
+    }
+
+    # Construct aggregate tables
+    for data_period in "1996-2003", "1996-2011", "2004-2011":
+        for table_type, table_no in zip(
             TABLE_TYPES, (CONC_TABLE_ALL, CNT_TABLE_ALL), strict=True
         ):
-            _invdata_sub_tabletype = _invdata[_data_period][_table_type]
+            invdata_sub_tabletype = invdata[data_period][table_type]
 
-            _aggr_tables_list = [
-                _t
-                for _t in _invdata["1996-2003"][_table_type]
+            aggr_tables_list = [
+                t_
+                for t_ in invdata["1996-2003"][table_type]
                 if re.sub(
-                    r"\W", "", _invdata["1996-2003"][_table_type][_t].industry_group
+                    r"\W", "", invdata["1996-2003"][table_type][t_].industry_group
                 )
-                not in _industry_exclusion_list
+                not in industry_exclusion_list
             ]
 
-            _invdata_sub_tabletype |= {
-                _table_no.replace(".1", ".X"): _invdata_build_aggregate_table(
-                    _invdata_sub_tabletype, _aggr_tables_list
+            invdata_sub_tabletype |= {
+                table_no.replace(".1", ".X"): invdata_build_aggregate_table(
+                    invdata_sub_tabletype, aggr_tables_list
                 )
             }
 
-    _ = INVDATA_ARCHIVE_PATH.write_bytes(msgpack.packb(_invdata))
+    retval: INVData = _mappingproxy_from_mapping(invdata)
+    with (
+        ZipFile(_archive_path, "w", compression=ZIP_DEFLATED) as _yzh,
+        _yzh.open(f"{_archive_path.stem}.yaml", "w") as _yfh,
+    ):
+        this_yaml.dump(retval, _yfh)
 
-    return MappingProxyType(_invdata)
+    return retval
 
 
-def _construct_no_evidence_data(_invdata: _INVData_in, _data_period: str, /) -> None:
-    _invdata_ind_grp = "All Markets"
-    _table_nos_map = dict(
+def _construct_no_evidence_data(_invdata: INVData_in_, _data_period: str, /) -> None:
+    invdata_ind_grp = "All Markets"
+    table_nos_map = dict(
         zip(
             (
                 "No Entry Evidence",
@@ -240,28 +234,28 @@ def _construct_no_evidence_data(_invdata: _INVData_in, _data_period: str, /) ->
             strict=True,
         )
     )
-    for _invdata_evid_cond in (
+    for invdata_evid_cond in (
         "No Entry Evidence",
         "No Evidence on Customer Complaints",
         "No Evidence on Hot Documents",
     ):
-        for _stats_grp in ("ByHHIandDelta", "ByFirmCount"):
-            _invdata_sub_evid_cond_conc = _invdata[_data_period][_stats_grp]
+        for stats_grp in ("ByHHIandDelta", "ByFirmCount"):
+            invdata_sub_evid_cond_conc = _invdata[_data_period][stats_grp]
 
-            _dtn = _table_nos_map[_invdata_evid_cond]["ByHHIandDelta"]
-            _stn0 = "Table 4.1" if _stats_grp == "ByFirmCount" else "Table 3.1"
-            _stn1, _stn2 = (_dtn.replace(".X", f".{_i}") for _i in ("1", "2"))
+            dtn = table_nos_map[invdata_evid_cond]["ByHHIandDelta"]
+            stn0 = "Table 4.1" if stats_grp == "ByFirmCount" else "Table 3.1"
+            stn1, stn2 = (dtn.replace(".X", f".{_i}") for _i in ("1", "2"))
 
-            _invdata_sub_evid_cond_conc |= {
-                _dtn: INVTableData(
-                    _invdata_ind_grp,
-                    _invdata_evid_cond,
+            invdata_sub_evid_cond_conc |= {
+                dtn: INVTableData(
+                    invdata_ind_grp,
+                    invdata_evid_cond,
                     np.column_stack((
-                        _invdata_sub_evid_cond_conc[_stn0].data_array[:, :2],
+                        invdata_sub_evid_cond_conc[stn0].data_array[:, :2],
                         (
-                            _invdata_sub_evid_cond_conc[_stn0].data_array[:, 2:]
-                            - _invdata_sub_evid_cond_conc[_stn1].data_array[:, 2:]
-                            - _invdata_sub_evid_cond_conc[_stn2].data_array[:, 2:]
+                            invdata_sub_evid_cond_conc[stn0].data_array[:, 2:]
+                            - invdata_sub_evid_cond_conc[stn1].data_array[:, 2:]
+                            - invdata_sub_evid_cond_conc[stn2].data_array[:, 2:]
                         ),
                     )),
                 )
@@ -275,44 +269,44 @@ def _construct_new_period_data(
     *,
     flag_backward_compatibility: bool = False,
 ) -> dict[str, dict[str, INVTableData]]:
-    _cuml_period = "1996-{}".format(int(_data_period.split("-")[1]))
-    if _cuml_period != "1996-2011":
+    cuml_period = f"1996-{_data_period.split('-')[1]}"
+    if cuml_period != "1996-2011":
         raise ValueError('Expected cumulative period, "1996-2011"')
 
-    _invdata_cuml = _invdata[_cuml_period]
+    invdata_cuml = _invdata[cuml_period]
 
-    _base_period = "1996-{}".format(int(_data_period.split("-")[0]) - 1)
-    _invdata_base = _invdata[_base_period]
+    base_period = "1996-{}".format(int(_data_period.split("-")[0]) - 1)
+    invdata_base = _invdata[base_period]
 
-    if tuple(_invdata_cuml.keys()) != TABLE_TYPES:
+    if tuple(invdata_cuml.keys()) != TABLE_TYPES:
         raise ValueError("Source data does not include the expected groups of tables.")
 
-    _invdata_bld = {}
-    for _table_type in TABLE_TYPES:
-        _data_typesubdict = {}
-        for _table_no in _invdata_cuml[_table_type]:
-            _invdata_cuml_sub_table = _invdata_cuml[_table_type][_table_no]
-            _invdata_ind_group, _invdata_evid_cond, _invdata_cuml_array = (
-                _invdata_cuml_sub_table.industry_group,
-                _invdata_cuml_sub_table.additional_evidence,
-                _invdata_cuml_sub_table.data_array,
+    invdata_bld = {}
+    for table_type in TABLE_TYPES:
+        data_typesubdict = {}
+        for table_no in invdata_cuml[table_type]:
+            invdata_cuml_sub_table = invdata_cuml[table_type][table_no]
+            invdata_ind_group, invdata_evid_cond, invdata_cuml_array = (
+                invdata_cuml_sub_table.industry_group,
+                invdata_cuml_sub_table.additional_evidence,
+                invdata_cuml_sub_table.data_array,
             )
 
-            _invdata_base_sub_table = _invdata_base[_table_type].get(
-                _table_no, INVTableData("", "", EMPTY_ARRAYINT)
+            invdata_base_sub_table = invdata_base[table_type].get(
+                table_no, INVTableData("", "", EMPTY_ARRAYINT)
             )
 
-            (_invdata_base_ind_group, _invdata_base_evid_cond, _invdata_base_array) = (
-                getattr(_invdata_base_sub_table, _a)
+            (invdata_base_ind_group, invdata_base_evid_cond, invdata_base_array) = (
+                getattr(invdata_base_sub_table, _a)
                 for _a in ("industry_group", "additional_evidence", "data_array")
             )
 
             # Some tables can't be constructed due to inconsistencies in the data
             # across time periods
             if (
-                (_data_period != "2004-2011" and _invdata_ind_group != "All Markets")
-                or (_invdata_ind_group in ('"Other" Markets', "Industries in Common"))
-                or (_invdata_base_ind_group in ('"Other" Markets', ""))
+                (_data_period != "2004-2011" and invdata_ind_group != "All Markets")
+                or (invdata_ind_group in {'"Other" Markets', "Industries in Common"})
+                or (invdata_base_ind_group in {'"Other" Markets', ""})
             ):
                 continue
 
@@ -328,31 +322,29 @@ def _construct_new_period_data(
                 # The number of "revisions" applied below, for enforcing consistency,
                 # is sufficiently small as to be unlikely to substantially impact
                 # results from analysis of the data.
-                _invdata_cuml_array_stack = []
-                _invdata_base_array_stack = []
-
-                for _data_period_detail in _invdata:
-                    _pd_start, _pd_end = (
-                        int(g) for g in _data_period_detail.split("-")
-                    )
-                    if _pd_start == 1996:
-                        _invdata_cuml_array_stack += [
-                            _invdata[_data_period_detail][_table_type][
-                                _table_no
+                invdata_cuml_array_stack = []
+                invdata_base_array_stack = []
+
+                for data_period_detail in _invdata:
+                    pd_start, pd_end = (int(g) for g in data_period_detail.split("-"))
+                    if pd_start == 1996:
+                        invdata_cuml_array_stack += [
+                            _invdata[data_period_detail][table_type][
+                                table_no
                             ].data_array[:, -3:-1]
                         ]
-                    if _pd_start == 1996 and _pd_end < int(_data_period.split("-")[0]):
-                        _invdata_base_array_stack += [
-                            _invdata[_data_period_detail][_table_type][
-                                _table_no
+                    if pd_start == 1996 and pd_end < int(_data_period.split("-")[0]):
+                        invdata_base_array_stack += [
+                            _invdata[data_period_detail][table_type][
+                                table_no
                             ].data_array[:, -3:-1]
                         ]
-                _invdata_cuml_array_enfcls, _invdata_base_array_enfcls = (
+                invdata_cuml_array_enfcls, invdata_base_array_enfcls = (
                     np.stack(_f).max(axis=0)
-                    for _f in (_invdata_cuml_array_stack, _invdata_base_array_stack)
+                    for _f in (invdata_cuml_array_stack, invdata_base_array_stack)
                 )
-                _invdata_array_bld_enfcls = (
-                    _invdata_cuml_array_enfcls - _invdata_base_array_enfcls
+                invdata_array_bld_enfcls = (
+                    invdata_cuml_array_enfcls - invdata_base_array_enfcls
                 )
             else:
                 # Consistency here means that the most recent data are considered
@@ -369,59 +361,59 @@ def _construct_new_period_data(
                 # backward compatible due to minor variation in (applying) the criteria
                 # for inclusion, as well as industry coding, undertaken to maintain
                 # transparency on the enforcement process.
-                _invdata_array_bld_enfcls = (
-                    _invdata_cuml_array[:, -3:-1] - _invdata_base_array[:, -3:-1]  # type: ignore
+                invdata_array_bld_enfcls = (
+                    invdata_cuml_array[:, -3:-1] - invdata_base_array[:, -3:-1]
                 )
 
                 # # // spellchecker: disable
                 # To examine the number of corrected values per table,  // spellchecker: disable
                 # uncomment the statements below
-                # _invdata_array_bld_tbc = where(
-                #   _invdata_array_bld_enfcls < 0, _invdata_array_bld_enfcls, 0
+                # invdata_array_bld_tbc = where(
+                #   invdata_array_bld_enfcls < 0, invdata_array_bld_enfcls, 0
                 # )
                 # if np.einsum('ij->', invdata_array_bld_tbc):
                 #     print(
-                #       f"{_data_period}, {_table_no}, {_invdata_ind_group}:",
+                #       f"{_data_period}, {_table_no}, {invdata_ind_group}:",
                 #       abs(np.einsum('ij->', invdata_array_bld_tbc))
                 #       )
                 # #  // spellchecker: disable
 
                 # Enforce non-negativity
-                _invdata_array_bld_enfcls = np.stack((
-                    _invdata_array_bld_enfcls,
-                    np.zeros_like(_invdata_array_bld_enfcls),
+                invdata_array_bld_enfcls = np.stack((
+                    invdata_array_bld_enfcls,
+                    np.zeros_like(invdata_array_bld_enfcls),
                 )).max(axis=0)
 
-            _invdata_array_bld = np.column_stack((
-                _invdata_cuml_array[:, :-3],
-                _invdata_array_bld_enfcls,
-                np.einsum("ij->i", _invdata_array_bld_enfcls),
+            invdata_array_bld = np.column_stack((
+                invdata_cuml_array[:, :-3],
+                invdata_array_bld_enfcls,
+                np.einsum("ij->i", invdata_array_bld_enfcls),
             ))
 
-            _data_typesubdict[_table_no] = INVTableData(
-                _invdata_ind_group, _invdata_evid_cond, _invdata_array_bld
+            data_typesubdict[table_no] = INVTableData(
+                invdata_ind_group, invdata_evid_cond, invdata_array_bld
             )
-            del _invdata_ind_group, _invdata_evid_cond, _invdata_cuml_array
-            del _invdata_base_ind_group, _invdata_base_evid_cond, _invdata_base_array
-            del _invdata_array_bld
-        _invdata_bld[_table_type] = _data_typesubdict
-    return _invdata_bld
+            del invdata_ind_group, invdata_evid_cond, invdata_cuml_array
+            del invdata_base_ind_group, invdata_base_evid_cond, invdata_base_array
+            del invdata_array_bld
+        invdata_bld[table_type] = data_typesubdict
+    return invdata_bld
 
 
-def _invdata_build_aggregate_table(
+def invdata_build_aggregate_table(
     _data_typesub: dict[str, INVTableData], _aggr_table_list: Sequence[str]
 ) -> INVTableData:
-    _hdr_table_no = _aggr_table_list[0]
+    hdr_table_no = _aggr_table_list[0]
 
     return INVTableData(
         "Industries in Common",
         "Unrestricted on additional evidence",
         np.column_stack((
-            _data_typesub[_hdr_table_no].data_array[:, :-3],
+            _data_typesub[hdr_table_no].data_array[:, :-3],
             np.einsum(
                 "ijk->jk",
                 np.stack([
-                    (_data_typesub[_t]).data_array[:, -3:] for _t in _aggr_table_list
+                    (_data_typesub[t_]).data_array[:, -3:] for t_ in _aggr_table_list
                 ]),
             ),
         )),
@@ -439,294 +431,295 @@ def _parse_invdata() -> INVData:
         by range of HHI and ∆HHI.
 
     """
-    raise ValueError(
-        "This function is defined here as documentation.\n"
-        "NOTE: License for `pymupdf`, upon which this function depends,"
-        " may be incompatible with the MIT license,"
-        " under which this pacakge is distributed."
-        " Making this fumction operable requires the user to modify"
-        " the source code as well as to install an additional package"
-        " not distributed with this package or included in its dependencies."
-    )
-    import fitz  # type: ignore
-
-    _invdata_docnames = _download_invdata(FTCDATA_DIR)
-
-    _invdata: dict[str, dict[str, dict[str, INVTableData]]] = {}
-
-    for _invdata_docname in _invdata_docnames:
-        _invdata_pdf_path = FTCDATA_DIR.joinpath(_invdata_docname)
-
-        _invdata_fitz = fitz.open(_invdata_pdf_path)
-        _invdata_meta = _invdata_fitz.metadata
-        if _invdata_meta["title"] == " ":
-            _invdata_meta["title"] = ", ".join((
+    # raise ValueError(
+    #     "This function is defined here as documentation.\n"
+    #     "NOTE: License for `pymupdf`, upon which this function depends,"
+    #     " may be incompatible with the MIT license,"
+    #     " under which this pacakge is distributed."
+    #     " Making this fumction operable requires the user to modify"
+    #     " the source code as well as to install an additional package"
+    #     " not distributed with this package or identified as a requirement."
+    # )
+    import pymupdf  # type: ignore  # noqa: PLC0415
+
+    invdata_docnames = _download_invdata(FTCDATA_DIR)
+
+    invdata: INVData_in_ = {}
+
+    for invdata_docname in invdata_docnames:
+        invdata_pdf_path = FTCDATA_DIR.joinpath(invdata_docname)
+
+        invdata_doc = pymupdf.open(invdata_pdf_path)
+        invdata_meta = invdata_doc.metadata
+        if invdata_meta["title"] == " ":
+            invdata_meta["title"] = ", ".join((
                 "Horizontal Merger Investigation Data",
                 "Fiscal Years",
                 "1996-2005",
             ))
 
-        _data_period = re.findall(r"(\d{4}) *(-) *(\d{4})", _invdata_meta["title"])[0]
-        _data_period = "".join(_data_period)
+        data_period = "".join(  # line-break here for readability
+            re.findall(r"(\d{4}) *(-) *(\d{4})", invdata_meta["title"])[0]
+        )
 
         # Initialize containers for parsed data
-        _invdata[_data_period] = {k: {} for k in TABLE_TYPES}
+        invdata[data_period] = {k: {} for k in TABLE_TYPES}
 
-        for _pdf_pg in _invdata_fitz.pages():
-            _doc_pg_blocks = _pdf_pg.get_text("blocks", sort=False)
+        for pdf_pg in invdata_doc.pages():
+            doc_pg_blocks = pdf_pg.get_text("blocks", sort=False)
             # Across all published reports of FTC investigations data,
             #   sorting lines (PDF page blocks) by the lower coordinates
             #   and then the left coordinates is most effective for
             #   ordering table rows in top-to-bottom order; this doesn't
             #   work for the 1996-2005 data, however, so we resort later
-            _doc_pg_blocks = sorted([
+            doc_pg_blocks = sorted([
                 (f"{_f[3]:03.0f}{_f[0]:03.0f}{_f[1]:03.0f}{_f[2]:03.0f}", *_f)
-                for _f in _doc_pg_blocks
+                for _f in doc_pg_blocks
                 if _f[-1] == 0
             ])
 
-            _data_blocks: list[tuple[str]] = [("",)]
+            data_blocks: list[tuple[str]] = [("",)]
             # Pages layouts not the same in all reports
-            _pg_hdr_strings = (
+            pg_hdr_strings = (
                 "FEDERAL TRADE COMMISSION",
                 "HORIZONTAL MERGER INVESTIGATION DATA: FISCAL YEARS 1996 - 2011",
             )
-            if len(_doc_pg_blocks) > 4:
-                _tnum: re.match = None
-                for _blk_idx, _pg_blk in enumerate(_doc_pg_blocks):
-                    if _tnum := TABLE_NO_RE.fullmatch(_pg_blk[-3].strip()):
-                        _data_blocks = [
-                            _b
-                            for _b in _doc_pg_blocks
-                            if not _b[-3].startswith(_pg_hdr_strings)
+            if len(doc_pg_blocks) > 4:
+                tnum = None
+                for _pg_blk in doc_pg_blocks:
+                    if tnum := TABLE_NO_RE.fullmatch(_pg_blk[-3].strip()):
+                        data_blocks = [
+                            b_
+                            for b_ in doc_pg_blocks
+                            if not b_[-3].startswith(pg_hdr_strings)
                             and (
-                                _b[-3].strip()
-                                not in ("Significant Competitors", "Post Merger HHI")
+                                b_[-3].strip()
+                                not in {"Significant Competitors", "Post Merger HHI"}
                             )
-                            and not re.fullmatch(r"\d+", _b[-3].strip())
+                            and not re.fullmatch(r"\d+", b_[-3].strip())
                         ]
                         break
-                if not _tnum:
+                if not tnum:
                     continue
-                del _tnum
+                del tnum
             else:
                 continue
 
-            _parse_page_blocks(_invdata, _data_period, _data_blocks)
+            _parse_page_blocks(invdata, data_period, data_blocks)
 
-        _invdata_fitz.close()
+        invdata_doc.close()
 
-    return MappingProxyType(_invdata)
+    return _mappingproxy_from_mapping(invdata)
 
 
 def _parse_page_blocks(
-    _invdata: _INVData_in, _data_period: str, _doc_pg_blocks: Sequence[Sequence[Any]], /
+    _invdata: INVData_in_, _data_period: str, _doc_pg_blocks: Sequence[Sequence[Any]], /
 ) -> None:
     if _data_period != "1996-2011":
         _parse_table_blocks(_invdata, _data_period, _doc_pg_blocks)
     else:
-        _test_list = [
+        test_list = [
             (g, f[-3].strip())
             for g, f in enumerate(_doc_pg_blocks)
             if TABLE_NO_RE.fullmatch(f[-3].strip())
         ]
         # In the 1996-2011 report, there are 2 tables per page
-        if len(_test_list) == 1:
-            _table_a_blocks = _doc_pg_blocks
-            _table_b_blocks: Sequence[Sequence[Any]] = []
+        if len(test_list) == 1:
+            table_a_blocks = _doc_pg_blocks
+            table_b_blocks: Sequence[Sequence[Any]] = []
         else:
-            _table_a_blocks, _table_b_blocks = (
-                _doc_pg_blocks[_test_list[0][0] : _test_list[1][0]],
-                _doc_pg_blocks[_test_list[1][0] :],
+            table_a_blocks, table_b_blocks = (
+                _doc_pg_blocks[test_list[0][0] : test_list[1][0]],
+                _doc_pg_blocks[test_list[1][0] :],
             )
 
-        for _table_i_blocks in _table_a_blocks, _table_b_blocks:
-            if not _table_i_blocks:
+        for table_i_blocks in table_a_blocks, table_b_blocks:
+            if not table_i_blocks:
                 continue
-            _parse_table_blocks(_invdata, _data_period, _table_i_blocks)
+            _parse_table_blocks(_invdata, _data_period, table_i_blocks)
 
 
 def _parse_table_blocks(
-    _invdata: _INVData_in, _data_period: str, _table_blocks: Sequence[Sequence[str]], /
+    _invdata: INVData_in_, _data_period: str, _table_blocks: Sequence[Sequence[str]], /
 ) -> None:
-    _invdata_evid_cond = "Unrestricted on additional evidence"
-    _table_num, _table_ser, _table_type = _identify_table_type(
+    invdata_evid_cond = "Unrestricted on additional evidence"
+    table_num, table_ser, table_type = _identify_table_type(
         _table_blocks[0][-3].strip()
     )
 
     if _data_period == "1996-2011":
-        _invdata_ind_group = (
+        invdata_ind_group = (
             _table_blocks[1][-3].split("\n")[1]
-            if _table_num == "Table 4.8"
+            if table_num == "Table 4.8"
             else _table_blocks[2][-3].split("\n")[0]
         )
 
-        if _table_ser > 4:
-            _invdata_evid_cond = (
+        if table_ser > 4:
+            invdata_evid_cond = (
                 _table_blocks[2][-3].split("\n")[1]
-                if _table_ser in (9, 10)
+                if table_ser in {9, 10}
                 else _table_blocks[3][-3].strip()
             )
 
     elif _data_period == "1996-2005":
         _table_blocks = sorted(_table_blocks, key=itemgetter(6))
 
-        _invdata_ind_group = _table_blocks[3][-3].strip()
-        if _table_ser > 4:
-            _invdata_evid_cond = _table_blocks[5][-3].strip()
+        invdata_ind_group = _table_blocks[3][-3].strip()
+        if table_ser > 4:
+            invdata_evid_cond = _table_blocks[5][-3].strip()
 
-    elif _table_ser % 2 == 0:
-        _invdata_ind_group = _table_blocks[1][-3].split("\n")[2]
-        if (_evid_cond_teststr := _table_blocks[2][-3].strip()) == "Outcome":
-            _invdata_evid_cond = "Unrestricted on additional evidence"
+    elif table_ser % 2 == 0:
+        invdata_ind_group = _table_blocks[1][-3].split("\n")[2]
+        if (evid_cond_teststr := _table_blocks[2][-3].strip()) == "Outcome":
+            invdata_evid_cond = "Unrestricted on additional evidence"
         else:
-            _invdata_evid_cond = _evid_cond_teststr
+            invdata_evid_cond = evid_cond_teststr
 
     elif _table_blocks[3][-3].startswith("FTC Horizontal Merger Investigations"):
-        _invdata_ind_group = _table_blocks[3][-3].split("\n")[2]
-        _invdata_evid_cond = "Unrestricted on additional evidence"
+        invdata_ind_group = _table_blocks[3][-3].split("\n")[2]
+        invdata_evid_cond = "Unrestricted on additional evidence"
 
     else:
         # print(_table_blocks)
-        _invdata_evid_cond = (
+        invdata_evid_cond = (
             _table_blocks[1][-3].strip()
-            if _table_ser == 9
+            if table_ser == 9
             else _table_blocks[3][-3].strip()
         )
-        _invdata_ind_group = _table_blocks[4][-3].split("\n")[2]
+        invdata_ind_group = _table_blocks[4][-3].split("\n")[2]
 
-    if _invdata_ind_group == "Pharmaceutical Markets":
-        _invdata_ind_group = "Pharmaceuticals Markets"
+    if invdata_ind_group == "Pharmaceutical Markets":
+        invdata_ind_group = "Pharmaceuticals Markets"
 
     process_table_func = (
         _process_table_blks_conc_type
-        if _table_type == TABLE_TYPES[0]
+        if table_type == TABLE_TYPES[0]
         else _process_table_blks_cnt_type
     )
 
-    _table_array = process_table_func(_table_blocks)
-    if not isinstance(_table_array, np.ndarray) or _table_array.dtype != np.int64:
-        print(_table_num)
+    table_array = process_table_func(_table_blocks)
+    if not isinstance(table_array, np.ndarray) or table_array.dtype != np.uint64:
+        print(table_num)
         print(_table_blocks)
         raise ValueError
 
-    _table_data = INVTableData(_invdata_ind_group, _invdata_evid_cond, _table_array)
-    _invdata[_data_period][_table_type] |= {_table_num: _table_data}
+    table_data = INVTableData(invdata_ind_group, invdata_evid_cond, table_array)
+    _invdata[_data_period][table_type] |= {table_num: table_data}
 
 
 def _identify_table_type(_tnstr: str = CONC_TABLE_ALL, /) -> tuple[str, int, str]:
-    _tnum = _tnstr.split(" ")[1]
-    _tsub = int(_tnum.split(".")[0])
-    return _tnstr, _tsub, TABLE_TYPES[(_tsub + 1) % 2]
+    tnum = _tnstr.split(" ")[1]
+    tsub = int(tnum.split(".")[0])
+    return _tnstr, tsub, TABLE_TYPES[(tsub + 1) % 2]
 
 
 def _process_table_blks_conc_type(
     _table_blocks: Sequence[Sequence[str]], /
 ) -> ArrayBIGINT:
-    _conc_row_pat = re.compile(r"((?:0|\d,\d{3}) (?:- \d+,\d{3}|\+)|TOTAL)")
-
-    _col_titles_array = tuple(CONC_DELTA_DICT.values())
-    _col_totals: ArrayBIGINT = np.zeros(len(_col_titles_array), np.int64)
-    _invdata_array: ArrayBIGINT = np.array(None)
-
-    for _tbl_blk in _table_blocks:
-        if _conc_row_pat.match(_blk_str := _tbl_blk[-3]):
-            _row_list: list[str] = _blk_str.strip().split("\n")
-            _row_title: str = _row_list.pop(0)
-            _row_key: int = CONC_HHI_DICT[_row_title]
-            _row_total = np.array(_row_list.pop().replace(",", "").split("/"), np.int64)
-            _row_array_list: list[list[int]] = []
-            while _row_list:
-                _enfd_val, _clsd_val = _row_list.pop(0).split("/")
-                _row_array_list += [
+    conc_row_pat = re.compile(r"((?:0|\d,\d{3}) (?:- \d+,\d{3}|\+)|TOTAL)")
+
+    col_titles_array = tuple(CONC_DELTA_DICT.values())
+    col_totals: ArrayBIGINT = np.zeros(len(col_titles_array), np.uint64)
+    invdata_array: ArrayBIGINT = np.array(None)
+
+    for tbl_blk in _table_blocks:
+        if conc_row_pat.match(_blk_str := tbl_blk[-3]):
+            row_list: list[str] = _blk_str.strip().split("\n")
+            row_title: str = row_list.pop(0)
+            row_key: int = CONC_HHI_DICT[row_title]
+            row_total = np.array(row_list.pop().replace(",", "").split("/"), np.uint64)
+            row_array_list: list[list[int]] = []
+            while row_list:
+                enfd_val, clsd_val = row_list.pop(0).split("/")
+                row_array_list += [
                     [
-                        _row_key,
-                        _col_titles_array[len(_row_array_list)],
-                        int(_enfd_val),
-                        int(_clsd_val),
-                        int(_enfd_val) + int(_clsd_val),
+                        row_key,
+                        col_titles_array[len(row_array_list)],
+                        int(enfd_val),
+                        int(clsd_val),
+                        int(enfd_val) + int(clsd_val),
                     ]
                 ]
-            _row_array = np.array(_row_array_list, np.int64)
+            row_array = np.array(row_array_list, np.uint64)
             # Check row totals
-            assert_array_equal(_row_total, np.einsum("ij->j", _row_array[:, 2:4]))
+            assert_array_equal(row_total, np.einsum("ij->j", row_array[:, 2:4]))
 
-            if _row_key == TTL_KEY:
-                _col_totals = _row_array
+            if row_key == TTL_KEY:
+                col_totals = row_array
             else:
-                _invdata_array = (
-                    np.vstack((_invdata_array, _row_array))
-                    if _invdata_array.shape
-                    else _row_array
+                invdata_array = (
+                    np.vstack((invdata_array, row_array))
+                    if invdata_array.shape
+                    else row_array
                 )
-            del _row_array, _row_array_list
+            del row_array, row_array_list
         else:
             continue
 
     # Check column totals
-    for _col_tot in _col_totals:
+    for _col_tot in col_totals:
         assert_array_equal(
             _col_tot[2:],
             np.einsum(
-                "ij->j", _invdata_array[_invdata_array[:, 1] == _col_tot[1]][:, 2:]
+                "ij->j", invdata_array[invdata_array[:, 1] == _col_tot[1]][:, 2:]
             ),
         )
 
-    return _invdata_array[
-        np.argsort(np.einsum("ij,ij->i", [[100, 1]], _invdata_array[:, :2]))
+    return invdata_array[
+        np.argsort(np.einsum("ij,ij->i", [[100, 1]], invdata_array[:, :2]))
     ]
 
 
 def _process_table_blks_cnt_type(
     _table_blocks: Sequence[Sequence[str]], /
 ) -> ArrayBIGINT:
-    _cnt_row_pat = re.compile(r"(\d+ (?:to \d+|\+)|TOTAL)")
+    cnt_row_pat = re.compile(r"(\d+ (?:to \d+|\+)|TOTAL)")
 
-    _invdata_array: ArrayBIGINT = np.array(None)
-    _col_totals: ArrayBIGINT = np.zeros(3, np.int64)  # "enforced", "closed", "total"
+    invdata_array: ArrayBIGINT = np.array(None)
+    col_totals: ArrayBIGINT = np.zeros(3, np.uint64)  # "enforced", "closed", "total"
 
     for _tbl_blk in _table_blocks:
-        if _cnt_row_pat.match(_blk_str := _tbl_blk[-3]):
-            _row_list_s = _blk_str.strip().replace(",", "").split("\n")
-            _row_list = np.array(
-                [CNT_FCOUNT_DICT[_row_list_s[0]], *_row_list_s[1:]], np.int64
+        if cnt_row_pat.match(_blk_str := _tbl_blk[-3]):
+            row_list_s = _blk_str.strip().replace(",", "").split("\n")
+            row_list = np.array(
+                [CNT_FCOUNT_DICT[row_list_s[0]], *row_list_s[1:]], np.uint64
             )
-            del _row_list_s
-            if _row_list[3] != _row_list[1] + _row_list[2]:
+            del row_list_s
+            if row_list[3] != row_list[1] + row_list[2]:
                 raise ValueError(
                     "Total number of investigations does not equal #enforced plus #closed."
                 )
-            if _row_list[0] == TTL_KEY:
-                _col_totals = _row_list
+            if row_list[0] == TTL_KEY:
+                col_totals = row_list
             else:
-                _invdata_array = (
-                    np.vstack((_invdata_array, _row_list))
-                    if _invdata_array.shape
-                    else _row_list
+                invdata_array = (
+                    np.vstack((invdata_array, row_list))
+                    if invdata_array.shape
+                    else row_list
                 )
         else:
             continue
 
     if not np.array_equal(
-        np.array([int(f) for f in _col_totals[1:]], np.int64),
-        np.einsum("ij->j", _invdata_array[:, 1:]),
+        np.array(list(col_totals[1:]), np.uint64),
+        np.einsum("ij->j", invdata_array[:, 1:]),
     ):
         raise ValueError("Column totals don't compute.")
 
-    return _invdata_array[np.argsort(_invdata_array[:, 0])]
+    return invdata_array[np.argsort(invdata_array[:, 0])]
 
 
 def _download_invdata(_dl_path: Path = FTCDATA_DIR) -> tuple[str, ...]:
     if not _dl_path.is_dir():
         _dl_path.mkdir(parents=True)
 
-    _invdata_homepage_urls = (
+    invdata_homepage_urls = (
         "https://www.ftc.gov/reports/horizontal-merger-investigation-data-fiscal-years-1996-2003",
         "https://www.ftc.gov/reports/horizontal-merger-investigation-data-fiscal-years-1996-2005-0",
         "https://www.ftc.gov/reports/horizontal-merger-investigation-data-fiscal-years-1996-2007-0",
         "https://www.ftc.gov/reports/horizontal-merger-investigation-data-fiscal-years-1996-2011",
     )
-    _invdata_docnames = (
+    invdata_docnames = (
         "040831horizmergersdata96-03.pdf",
         "p035603horizmergerinvestigationdata1996-2005.pdf",
         "081201hsrmergerdata.pdf",
@@ -734,40 +727,40 @@ def _download_invdata(_dl_path: Path = FTCDATA_DIR) -> tuple[str, ...]:
     )
 
     if all(
-        _dl_path.joinpath(_invdata_docname).is_file()
-        for _invdata_docname in _invdata_docnames
+        _dl_path.joinpath(invdata_docname).is_file()
+        for invdata_docname in invdata_docnames
     ):
-        return _invdata_docnames
-
-    _invdata_docnames_dl: tuple[str, ...] = ()
-    _u3pm = urllib3.PoolManager()
-    _chunk_size = 1024 * 1024
-    for _invdata_homepage_url in _invdata_homepage_urls:
-        with _u3pm.request(
-            "GET", _invdata_homepage_url, preload_content=False
+        return invdata_docnames
+
+    invdata_docnames_dl: tuple[str, ...] = ()
+    u3pm = urllib3.PoolManager()
+    chunk_size_ = 1024 * 1024
+    for invdata_homepage_url in invdata_homepage_urls:
+        with u3pm.request(
+            "GET", invdata_homepage_url, preload_content=False
         ) as _u3handle:
-            _invdata_soup = BeautifulSoup(_u3handle.data, "html.parser")
-            _invdata_attrs = [
+            invdata_soup = BeautifulSoup(_u3handle.data, "html.parser")
+            invdata_attrs = [
                 (_g.get("title", ""), _g.get("href", ""))
-                for _g in _invdata_soup.find_all("a")
+                for _g in invdata_soup.find_all("a")
                 if _g.get("title", "") and _g.get("href", "").endswith(".pdf")
             ]
-        for _invdata_attr in _invdata_attrs:
-            _invdata_docname, _invdata_link = _invdata_attr
-            _invdata_docnames_dl += (_invdata_docname,)
+        for invdata_attr in invdata_attrs:
+            invdata_docname, invdata_link = invdata_attr
+            invdata_docnames_dl += (invdata_docname,)
             with (
-                _u3pm.request(
-                    "GET", f"https://www.ftc.gov/{_invdata_link}", preload_content=False
+                u3pm.request(
+                    "GET", f"https://www.ftc.gov/{invdata_link}", preload_content=False
                 ) as _urlopen_handle,
-                _dl_path.joinpath(_invdata_docname).open("wb") as _invdata_fh,
+                _dl_path.joinpath(invdata_docname).open("wb") as invdata_fh,
             ):
                 while True:
-                    _data = _urlopen_handle.read(_chunk_size)
-                    if not _data:
+                    data = _urlopen_handle.read(chunk_size_)
+                    if not data:
                         break
-                    _invdata_fh.write(_data)
+                    invdata_fh.write(data)
 
-    return _invdata_docnames_dl
+    return invdata_docnames_dl
 
 
 if __name__ == "__main__":