# Attribution allowlist — every rendered clause that credits an authority or
# asserts a rank, across the four disjoint scenarios in
# tests/test_invariant_output.py, in all four formats.
#
# Format:  KIND | citation | normalised clause (digits collapsed to N)
#
#   CITED    a document, and a year / round / section. The gate additionally
#            requires a digit in the citation, so "CDFI Fund guidance" alone
#            will not pass — a year, section or number must appear.
#   HOUSE    this tool's own judgement. The gate does NOT require a digit, but
#            the RENDERED text must say so on its face; an allowlist entry
#            claiming HOUSE while the document reads as a Fund figure is the
#            defect, not the fix.
#   PLACEHO  inside bracketed [CDE TO COMPLETE] / [NARRATIVE PLACEHOLDER]
#            guidance — marked on its face as instruction, not assertion
#   LABEL    a bare column or row label containing a trigger word
#
# NO ENTRY IS MARKED "PROVENANCE UNVERIFIED" ANY MORE, AND THAT IS THE POINT.
# Six were, in the first version of this file. A primary-source pass on those
# six established that the publication they cited — a "CDFI Fund NMTC Program
# Annual Report, FY2018-FY2023" — DOES NOT EXIST, that the Fund publishes no
# jobs-per-QEI figure in any denominator, and that no distribution exists to
# make 20.0 a "top quartile". The rendered text was relabelled as this tool's
# own screening bands, so those six clauses no longer render at all and their
# entries are gone. They were not re-cited and not downgraded to HOUSE — the
# claims were withdrawn. See nmtcapp/data/schema.py.
#
# NO ENTRY READS "DOCUMENT RETRIEVABLE, CONTENT NOT LOCATED" ANY MORE EITHER.
# Three did, all three for the distress-level definitions. That queue is now
# closed, and closing it found a defect worse than the missing citation.
#
# The rendered text read:
#
#     "Deep distress = poverty rate >30% or unemployment >1.5x national
#      average. Severe distress = LIC plus additional qualifying factors."
#
# Those thresholds are real. They are the CDFI Fund's definition of SEVERE
# distress, printed under the label DEEP. Verified by opening the Fund's own
# NMTC LIC Eligibility workbook — the .xlsb this package downloads and loads —
# and reading columns 14 and 15 verbatim:
#
#     col 14: "Severe distress=LIC AND (Poverty>30%; MFI<=60%;Unemployment>=1.5)"
#     col 15: "Deep distress=LIC AND (Poverty>40%; MFI<=40%;Unemployment>=2.5)"
#
# So the tool had the two tiers' thresholds swapped, dropped the median-family-
# income limb from both, dropped the "AND LIC" term from both, and attributed
# the result to "CDFI Fund NMTC Program guidance and the NMTC Allocation
# Application Review Process" — neither of which carries that wording. A CDE
# reading the old note was told a 32%-poverty tract is "deep distress"; the
# Fund's deep-distress bar is 40%.
#
# The classification DATA was never wrong — distress_level comes from these same
# workbook columns via nmtc-mapper. Only the printed definition was. That is its
# own lesson: correct data under an incorrect legend is still a wrong document.
#
# Cite the file you actually load. It is more retrievable than the prose
# document describing it, and it is the thing the numbers came from.

PLACEHO | Wrapped continuation of the [NARRATIVE PLACEHOLDER] block. Reclassified from CITED in 1.2.0: it is guidance addressed to the CDE about what to write, not a figure, and the claim about reviewer emphasis was never located in a Fund document. N2 misses it only because the PDF extractor wraps the line away from its opening bracket. | cdfi fund reviewers score on specificity, evidence, and alignment with
PLACEHO | Wrapped tail of INVESTOR_TABLE_NOTE, a [CDE TO COMPLETE] block. Names the Fund as the audience a CDE must defend a row to. | cdfi fund, and delete any row you cannot.] (see attachment: investor commitments table)
PLACEHO | Inside the has_prior_reporting_issues [CDE TO COMPLETE] block. A statement about the Fund's own recordkeeping, in guidance addressed to the CDE, not a figure in the CDE's voice. | do not omit this — the cdfi fund holds prior
LABEL | Column header in the Section B distress-documentation table. Names an ACS field; the value beneath it renders 'See ACS'. Claims nothing. | median family income
CITED | CDFI Fund NMTC LIC Eligibility workbook (NMTC_LIC_Eligibility_2016_2020.xlsb), sheet '2016-2020', column headers 14 and 15, quoted verbatim: 'Severe distress=LIC AND (Poverty>30%; MFI<=60%;Unemployment>=1.5)' and 'Deep distress=LIC AND (Poverty>40%; MFI<=40%;Unemployment>=2.5)'. Read directly off the .xlsb this tool downloads and loads, not off a secondary description. Replaces the three CONTENT NOT LOCATED entries this file used to carry. | distress levels (verbatim from the cdfi fund nmtc lic eligibility workbook,
CITED | CDFI Fund NMTC LIC Eligibility workbook (NMTC_LIC_Eligibility_2016_2020.xlsb), sheet '2016-2020', column headers 14 and 15, quoted verbatim: 'Severe distress=LIC AND (Poverty>30%; MFI<=60%;Unemployment>=1.5)' and 'Deep distress=LIC AND (Poverty>40%; MFI<=40%;Unemployment>=2.5)'. Read directly off the .xlsb this tool downloads and loads, not off a secondary description. Replaces the three CONTENT NOT LOCATED entries this file used to carry. | distress levels (verbatim from the cdfi fund nmtc lic eligibility workbook, N-N acs, columns N and N — the file this tool loads to classify every tract): severe distress = lic and (poverty>N%;
LABEL | 'tiers' here names the two levels the Fund's own workbook defines; the clause states a structural fact about both definitions (each requires LIC) and claims no rank. | both tiers additionally require the tract to be a low-income community.
CITED | CY 2024-2025 NMTC Program Review Process, 'Targeting Areas of Higher Distress (Question 25)': 'at least 85% of its QLICIs in specified areas of severe distress and/or areas characterized by multiple indicia of distress ... at least 20% of its QLICIs to "Deep Distress" areas'. Downloaded and text-extracted; the document contains no 'FAQ #79', which is what the two entries this replaces used to cite. | cdfi fund higher-distress commitment (cy N-N)
CITED | CY 2024-2025 NMTC Program Review Process, 'Targeting Areas of Higher Distress (Question 25)': 'at least 85% of its QLICIs in specified areas of severe distress and/or areas characterized by multiple indicia of distress ... at least 20% of its QLICIs to "Deep Distress" areas'. Downloaded and text-extracted; the document contains no 'FAQ #79', which is what the two entries this replaces used to cite. | cdfi fund higher-distress commitment (cy N-N):** N% of qlicis in areas of severe distress and/or multiple indicia of distress, and N% in deep distress areas (cy N-N nmtc program review process, targeting areas of higher distress, question N;
