Sources and rights

• https://download.cms.gov/nppes/NPI_Files.html
• https://download.cms.gov/nppes/NPPES_Data_Dissemination_October_2026_V2.zip
• https://www.cms.gov/medicare/regulations-guidance/administrative-simplification/data-dissemination
• https://www.cms.gov/about-cms/web-policies-important-links/web-policies/privacy
• https://www.nucc.org/index.php/code-sets-mainmenu-41/provider-taxonomy-mainmenu-40/csv-mainmenu-57

Bulk file: "Full Replacement Monthly NPI File" V.2 zip found on the CMS NPI_Files page (latest full monthly file, not weekly/deactivation), downloaded once via lib/http downloadFile, streamed with zip.ts + csv-parse; nothing extracted to disk. Main file npidata_pfile_*.csv: only Entity Type Code = 2 and Practice Location state = WV or MD (each office carries its own state); columns read by header name: NPI, Entity Type Code, organization legal business name, practice-location address/phone/fax, NPI deactivation/reactivation dates, taxonomy code + primary-switch (1-15). No Authorized Official or other person column is indexed. Solo-practice legal names (a person name, usually with credentials) are dropped by scripts/sources/s01-nppes.person.ts and denylisted in persons_dropped. Reviewed surname-brand solo practices and a few off-type organizations are dropped by NPI number (lists in s01-nppes.ts; the lists hold identifiers only). Endpoint file endpoint_pfile_*.csv: Endpoint Type = DIRECT, Affiliation Address State = the state of the NPI practice location (WV or MD), NPI in the mapped Type-2 set. Taxonomy -> agency_type table (primary taxonomy first, then other taxonomies in column order only if the primary is unmapped): hospital_records: 282N00000X 282NC2000X 282NC0060X 282NR1301X 282NW0100X 281P00000X 281PC2000X 282E00000X 283X00000X 283XC2000X 284300000X 282J00000X 286500000X 2865C1500X 2865M2000X 2865X1600X 261QC0050X 273100000X 273Y00000X 275N00000X; fqhc_site: 261QF0400X 261QC1500X; bh_facility: 283Q00000X 273R00000X 276400000X 261QM0801X 261QM0850X 261QM0855X 261QM2800X 261QR0405X 324500000X 3245S0500X 323P00000X 322D00000X 320800000X 251S00000X. Codes checked against the NUCC code set (current version) at run time; hospital-unit codes 273100000X/273Y00000X/275N00000X mapped to hospital_records and 273R00000X/276400000X to bh_facility by our decision. Rights pages checked: https://download.cms.gov/nppes/NPI_Files.html, https://www.cms.gov/medicare/regulations-guidance/administrative-simplification/data-dissemination, https://www.cms.gov/about-cms/web-policies-important-links/web-policies/privacy, and the Readme PDF inside the zip. None states a license, copyright or reuse restriction; the dissemination page says the data are FOIA-disclosable, required to be disclosed, no charge, no opt-out. NUCC code-set page https://www.nucc.org/index.php/code-sets-mainmenu-41/provider-taxonomy-mainmenu-40/csv-mainmenu-57 carries an AMA copyright and a license requirement for commercial use of the code set itself; this module uses NUCC only to verify code numbers at run time and stores our own short labels for the mapped codes, not NUCC definitions.

• https://gisportal.hrsa.gov/server/rest/services/HealthCareFacilities/HealthCareFacilities/MapServer/18?f=json
• https://gisportal.hrsa.gov/server/rest/services/HealthCareFacilities/HealthCareFacilities/MapServer?f=json
• https://gisportal.hrsa.gov/server/rest/services/HealthCareFacilities/HealthCareFacilities/MapServer/18/metadata
• https://gisportal.hrsa.gov/portal/sharing/rest/content/items/9b1abc90ef314de891cc526051057811?f=json
• https://gisportal.hrsa.gov/server/rest/services/HealthCareFacilities/HealthCareFacilities/MapServer/18/query?where=SITE_STATE_ABBR%20IN%20('WV'%2C'MD'%2C'OH')&returnCountOnly=true&f=json
• https://gisportal.hrsa.gov/server/rest/services/HealthCareFacilities/HealthCareFacilities/MapServer/18/query?where=SITE_STATE_ABBR%20IN%20('WV'%2C'MD'%2C'OH')&outFields=*&returnGeometry=false&orderByFields=OBJECTID&resultOffset=0&resultRecordCount=1000&f=json
• https://data.hrsa.gov/data/download
• https://www.hhs.gov/disclaimer.html
• https://www.hhs.gov/privacy/privacy-policy/index.html

ArcGIS REST layer 18 queried with where=SITE_STATE_ABBR IN ('WV','MD','OH'), outFields=*, returnGeometry=false, ordered by OBJECTID and paged by resultOffset/resultRecordCount (page size min(layer maxRecordCount, 1000)); the row count is cross-checked against a returnCountOnly query. Rights: layer description, layer copyrightText, service copyrightText, layer /metadata XML (empty abstract/credit) and the portal item (licenseInfo null) carry no terms; the only statement found is the HRSA Data Warehouse downloads page "Data Usage Terms & Conditions" entry for Health Center Service Delivery Sites (Usage limitations: None). Site footer links (HHS Disclaimers, Privacy Policy) are attempted each run; hhs.gov answered 403 to our descriptive-UA fetcher on 2026-10-19, so they were not read. Mapping: local_id hrsa:<BPHC_SITE_NUM>; office_name SITE_NM; address "SITE_ADDRESS, SITE_CITY, <state> SITE_ZIP_CD"; state SITE_STATE_ABBR; county LIST_BOX_COUNTY_NM via normalizeCountyIn for that state; website SITE_URL (bare domains get an https:// prefix; n/a placeholders dropped); parent_org null (layer has no health-center organization name; GRANT_NUM only); phone SITE_PHONE_NUM; fax SITE_FAX_NUM; ADMIN_PHONE_NUM -> phone with conditions "Administrative phone" (merged into the site phone row when identical); hours "Total weekly hours: <TOT_OPER_HR_PER_WEEK>" on site-phone routes only. source_url per row = per-site query on the layer. source_date = DW_RECORD_CREATE_DT_TXT (identical on every row) which matches the "Updated" date on the downloads page.

• https://data.cms.gov/provider-characteristics/hospitals-and-other-facilities/provider-of-services-file-quality-improvement-and-evaluation-system
• https://data.cms.gov/provider-characteristics/hospitals-and-other-facilities/provider-of-services-file-internet-quality-improvement-and-evaluation-system
• https://data.cms.gov/data.json
• https://data.cms.gov/sites/default/files/2023-07/0ca58d5d-7914-4532-b22d-41741d3e6151/P.QWB.POSQ.OTHER.LAYOUT.MAR23.pdf
• https://data.cms.gov/sites/default/files/2026-10/1fea27a3-7603-4fd2-bc2a-3e868934a9bf/iQIES%20POS%20Data%20Dictionary%20Oct2026.xlsx
• https://data.cms.gov/terms-of-use
• https://data.cms.gov/about
• https://www.usa.gov/government-works
• https://www.usa.gov/government-copyright
• https://www.cms.gov/about-cms/web-policies-notices
• https://www.cms.gov/about-cms/web-policies-notices/website-policies
• https://www.cms.gov/privacy
• https://data.cms.gov/data-api/v1/dataset/8ba0f9b4-9493-4aa0-9f82-44ea9468d1b5/data?filter[STATE_CD]=WV&filter[PGM_TRMNTN_CD]=00
• https://data.cms.gov/data-api/v1/dataset/8ba0f9b4-9493-4aa0-9f82-44ea9468d1b5/data/stats?filter[STATE_CD]=WV
• https://data.cms.gov/data-api/v1/dataset/086e48c4-87a6-4be1-8823-29e8da8f225b/data?filter[state_cd]=WV&filter[pgm_trmntn_cd]=00
• https://data.cms.gov/data-api/v1/dataset/086e48c4-87a6-4be1-8823-29e8da8f225b/data/stats?filter[state_cd]=WV
• https://data.cms.gov/data-api/v1/dataset/8ba0f9b4-9493-4aa0-9f82-44ea9468d1b5/data?filter[STATE_CD]=MD&filter[PGM_TRMNTN_CD]=00
• https://data.cms.gov/data-api/v1/dataset/8ba0f9b4-9493-4aa0-9f82-44ea9468d1b5/data/stats?filter[STATE_CD]=MD
• https://data.cms.gov/data-api/v1/dataset/086e48c4-87a6-4be1-8823-29e8da8f225b/data?filter[state_cd]=MD&filter[pgm_trmntn_cd]=00
• https://data.cms.gov/data-api/v1/dataset/086e48c4-87a6-4be1-8823-29e8da8f225b/data/stats?filter[state_cd]=MD
• https://data.cms.gov/data-api/v1/dataset/8ba0f9b4-9493-4aa0-9f82-44ea9468d1b5/data?filter[STATE_CD]=OH&filter[PGM_TRMNTN_CD]=00
• https://data.cms.gov/data-api/v1/dataset/8ba0f9b4-9493-4aa0-9f82-44ea9468d1b5/data/stats?filter[STATE_CD]=OH
• https://data.cms.gov/data-api/v1/dataset/086e48c4-87a6-4be1-8823-29e8da8f225b/data?filter[state_cd]=OH&filter[pgm_trmntn_cd]=00
• https://data.cms.gov/data-api/v1/dataset/086e48c4-87a6-4be1-8823-29e8da8f225b/data/stats?filter[state_cd]=OH

data.cms.gov dataset pages are JavaScript shells, so rows come from the data.cms.gov Data API (JSON), not HTML. Dataset ids are resolved from the data.json catalog: for each ticket landing page the vintage with the latest temporal endDate. Server-side filters state=WV or MD (one pull per state) and termination code=00 are applied and re-verified locally (trimmed value "00"); the API /data/stats found_rows is used as a completeness cross-check for the paged pull. Field names differ by file (QIES: UPPER_CASE, PRVDR_CTGRY_CD/PRVDR_CTGRY_SBTYP_CD; iQIES: lower_case, prvdr_type_id/prvdr_sbtyp_id). QIES category/subtype labels come from the QIES record layout PDF (hospital subtypes: 04 Psychiatric -> bh_facility). iQIES provider type ids 3,5,6,7,8,10,11,12,13,20 are labeled in the iQIES data dictionary; ids 4 (PRTF), 9 (RHC), 14 (CMHC) and 15 (FQHC) are NOT in that dictionary table and were identified from the file itself: FQHC-only column fed_fundd_fqhc_sw is populated only for id 15, RHC-only nrs_prctnr_cnt only for id 9, ids 14/4 carry CMHC-series (46xx) and PRTF-series (xxLnnn) provider numbers. Rights: the catalog distribution license is https://www.usa.gov/government-works. Checked: data.cms.gov /terms-of-use and /about (JS shell, no text), usa.gov government-works (redirects to government-copyright; generic notice that not every item on a federal site is a government work and that agency logos/endorsement are restricted), cms.gov/privacy (privacy policy only, no data-use terms), cms.gov web-policies-notices and website-policies (HTTP 404). No term restricting commercial use or scraping found.

• https://nces.ed.gov/ccd/data/zip/ccd_lea_029_2526_w_0a_050626.zip
• https://nces.ed.gov/ccd/files.asp
• https://nces.ed.gov/ccd/datatables/api/File/0/5/0/0/0/0
• https://nces.ed.gov/help/disclaimer.asp
• https://nces.ed.gov/about/public-access-research
• https://nces.ed.gov/help/privacy.asp
• https://ies.ed.gov/help/privacy.asp
• https://nces.ed.gov/ccd/ccddata.asp
• https://nces.ed.gov/ccd/aboutCCD.asp

Bulk file: the CCD Data File Tool (nces.ed.gov/ccd/files.asp, an Angular page backed by the public JSON endpoint nces.ed.gov/ccd/datatables/api/File/0/5/0/0/0/0) lists ccd_lea_029_2526_w_0a_050626.zip as "Local Education Agency (School District) Universe Survey Data", school year 2025 - 2026; downloaded once with lib/http downloadFile (robots.txt honoured) and the CSV read in-stream from the zip with yauzl + csv-parse (nothing extracted to disk). Rows with ST = WV kept (68; FIPST 54 and STATENAME WEST VIRGINIA give the same 68) and rows with ST = MD (FIPST 24, STATENAME MARYLAND). The file has no county field, so county is null for every office (LEA names are not parsed). Address = location (L*) street lines + city, state, ZIP[-ZIP4]; the mailing (M*) fields are used only if the location street lines are empty. Rights pages checked: nces.ed.gov/help/disclaimer.asp (redirects to nces.ed.gov/about/public-access-research, whose "Permission to Replicate Information" section carries the public-domain statement), nces.ed.gov/help/privacy.asp (same redirect), ies.ed.gov/help/privacy.asp (privacy and security policy; no copyright/use restriction found), nces.ed.gov/ccd/ccddata.asp and nces.ed.gov/ccd/aboutCCD.asp (no data-use restriction found). Caveat: the quoted statement is literally scoped to the ies.ed.gov website (NCES is an IES center); no clause restricting commercial use or scraping was found anywhere on the NCES CCD pages. The CCD file does not publish its own release date; the zip name token 050626 is not documented on the pages checked, so source_date uses the EFFECTIVE_DATE stated inside the file. Every WV office row carries website https://wveis.k12.wv.us/school-directory/ (a link to the WVEIS school directory; the host is blocked in lib/http.ts and is never fetched); the NCES WEBSITE field is not stored for WV. A Maryland office carries the NCES WEBSITE value when it is a plain http(s) URL, else null.

• https://dohs.wv.gov/field-offices?field_county_target_id=All&page=0
• https://dohs.wv.gov/offices/kanawha-county-dohs-office
• https://dohs.wv.gov/robots.txt
• https://www.wv.gov/policies/Pages/default.aspx
• https://www.wv.gov/policies/disclaimer
• https://www.wv.gov/policies/legal-notices
• https://www.wv.gov/policies/link-policy
• https://www.wv.gov/privacy-policy

HTML only (no API). Walked the paginated Drupal listing page=0.. until a page had no office cards, then fetched every office detail page via lib/http.ts. Per detail page: h1 title, services body, physical address field, phone/fax tel: links, hours field. Mailing addresses (separate field) are not stored. Rights pages checked: dohs.wv.gov footer link "Privacy, Security and Accessibility" -> wv.gov/policies (index), wv.gov/policies/disclaimer (a warranty/reliance disclaimer only: it disclaims accuracy and timeliness and states no reuse, copying or scraping terms, so no rights clause is recorded), wv.gov/policies/legal-notices (DMCA/trademark/warranty text only), wv.gov/policies/link-policy (linking rules only), wv.gov/privacy-policy (visitor privacy only), dohs.wv.gov/nondiscrimination-policies-and-plans, dohs.wv.gov/sitemap and dohs.wv.gov/robots.txt (no disallow for /field-offices or /offices). No DoHS-specific terms-of-use, copyright or data-use page exists in the site footer, nav or sitemap, and no page bars reuse, commercial use or scraping. agency_type is decided from the services text on each detail page, not the title: Family Assistance programs (SNAP/Medicaid/TANF) -> family_assistance; Child Protective Services/Foster Care/Adoption -> child_welfare; both -> two office rows with duplicated routes; neither -> child_support (every such page is a BCSE "Child Support" page, asserted). local_id is dohs:<county slug>:<agency_type> (county + office kind), never the page slug or URL.

• https://bss.wv.gov/media/20955/download
• https://bss.wv.gov/
• https://www.wv.gov/policies/Pages/default.aspx
• https://www.wv.gov/policies/disclaimer
• https://www.wv.gov/policies/legal-notices
• https://www.wv.gov/policies/link-policy
• https://www.wv.gov/privacy-policy
• https://www.wv.gov/policies/security

Rights: the wv.gov content disclaimer (https://www.wv.gov/policies/disclaimer) only disclaims warranty of accuracy and timeliness and states reliance is at the user's own risk; it states no reuse, copying or scraping terms, so no rights clause is recorded. Single-page PDF (bss.wv.gov/media/20955/download, served as "BSS Regional Managers List 7.2025.pdf", page stamp "July 10, 2025 Updated"), fetched via scripts/lib/http.ts and parsed with pdfjs item x/y positions: items are assigned to 4 columns by x, lines grouped by y, then an explicit block state machine (heading -> people -> address -> Phone/Fax) builds one block per county group. Rights pages checked: bss.wv.gov home footer (links only to wv.gov policies), wv.gov policies index, Content Disclaimer, Legal Notices, Link Policy, Privacy Policy, Security Policy. None states a reuse, commercial-use or scraping restriction; the Content Disclaimer and Legal Notices only disclaim accuracy/liability and the DMCA agent is for the wv.gov portal. No license or open-data statement found, hence public_unresolved. Region North / Region South / Commissioner header blocks are not district blocks and are not emitted.

• https://www.courtswv.gov/public-resources/court-information-county/county-map
• https://www.courtswv.gov/court-administration/probation-services/contacts
• https://www.courtswv.gov/terms-use
• https://www.courtswv.gov/public-resources/court-information-county/barbour_county
• https://www.courtswv.gov/public-resources/court-information-county/berkeley_county
• https://www.courtswv.gov/public-resources/court-information-county/boone_county
• https://www.courtswv.gov/public-resources/court-information-county/braxton_county
• https://www.courtswv.gov/public-resources/court-information-county/brooke_county
• https://www.courtswv.gov/public-resources/court-information-county/cabell_county
• https://www.courtswv.gov/public-resources/court-information-county/calhoun_county
• https://www.courtswv.gov/public-resources/court-information-county/clay_county
• https://www.courtswv.gov/public-resources/court-information-county/doddridge_county
• https://www.courtswv.gov/public-resources/court-information-county/fayette_county
• https://www.courtswv.gov/public-resources/court-information-county/gilmer_county
• https://www.courtswv.gov/public-resources/court-information-county/grant_county
• https://www.courtswv.gov/public-resources/court-information-county/greenbrier_county
• https://www.courtswv.gov/public-resources/court-information-county/hampshire_county
• https://www.courtswv.gov/public-resources/court-information-county/hancock_county
• https://www.courtswv.gov/public-resources/court-information-county/hardy_county
• https://www.courtswv.gov/public-resources/court-information-county/harrison_county
• https://www.courtswv.gov/public-resources/court-information-county/jackson_county
• https://www.courtswv.gov/public-resources/court-information-county/jefferson_county
• https://www.courtswv.gov/public-resources/court-information-county/kanawha_county
• https://www.courtswv.gov/public-resources/court-information-county/lewis_county
• https://www.courtswv.gov/public-resources/court-information-county/lincoln_county
• https://www.courtswv.gov/public-resources/court-information-county/logan_county
• https://www.courtswv.gov/public-resources/court-information-county/marion_county
• https://www.courtswv.gov/public-resources/court-information-county/marshall_county
• https://www.courtswv.gov/public-resources/court-information-county/mason_county
• https://www.courtswv.gov/public-resources/court-information-county/mcdowell_county
• https://www.courtswv.gov/public-resources/court-information-county/mercer_county
• https://www.courtswv.gov/public-resources/court-information-county/mineral_county
• https://www.courtswv.gov/public-resources/court-information-county/mingo_county
• https://www.courtswv.gov/public-resources/court-information-county/monongalia_county
• https://www.courtswv.gov/public-resources/court-information-county/monroe_county
• https://www.courtswv.gov/public-resources/court-information-county/morgan_county
• https://www.courtswv.gov/public-resources/court-information-county/nicholas_county
• https://www.courtswv.gov/public-resources/court-information-county/ohio_county
• https://www.courtswv.gov/public-resources/court-information-county/pendleton_county
• https://www.courtswv.gov/public-resources/court-information-county/pleasants_county
• https://www.courtswv.gov/public-resources/court-information-county/pocahontas_county
• https://www.courtswv.gov/public-resources/court-information-county/preston_county
• https://www.courtswv.gov/public-resources/court-information-county/putnam_county
• https://www.courtswv.gov/public-resources/court-information-county/raleigh_county
• https://www.courtswv.gov/public-resources/court-information-county/randolph_county
• https://www.courtswv.gov/public-resources/court-information-county/ritchie_county
• https://www.courtswv.gov/public-resources/court-information-county/roane_county
• https://www.courtswv.gov/public-resources/court-information-county/summers_county
• https://www.courtswv.gov/public-resources/court-information-county/taylor_county
• https://www.courtswv.gov/public-resources/court-information-county/tucker_county
• https://www.courtswv.gov/public-resources/court-information-county/tyler_county
• https://www.courtswv.gov/public-resources/court-information-county/upshur_county
• https://www.courtswv.gov/public-resources/court-information-county/wayne_county
• https://www.courtswv.gov/public-resources/court-information-county/webster_county
• https://www.courtswv.gov/public-resources/court-information-county/wetzel_county
• https://www.courtswv.gov/public-resources/court-information-county/wirt_county
• https://www.courtswv.gov/public-resources/court-information-county/wood_county
• https://www.courtswv.gov/public-resources/court-information-county/wyoming_county

HTML only (no API, no bulk file). County list from the county-map index (55 links); each county page parsed with cheerio (accordion sections Circuit Court / Family Court / Magistrate Court / Probation; only clerk, family court and probation entries are used). Probation circuit numbers parsed from the "Nth Circuit (counties)" accordion headings of the probation-services contacts page; its names, e-mails and phone numbers are ignored and recorded in persons_dropped. Rights pages checked: county page footer (copyright line, quoted as the clause), /terms-use (Security Notice, General Disclaimer, Privacy Notice: no warranty/liability for accuracy or completeness; no reuse, commercial-use or scraping restriction stated), robots.txt (no Disallow touching /public-resources/ or /court-administration/). No separate copyright or data-use page found; the footer links only Terms of Use and Accessibility Statement. Mapping decisions: family-court and probation blocks are folded together by physical address (town + street, suite/room ignored) so the same office listed on several county pages becomes one office row; one route row per (office, channel, value, page). The office local_id is court:<county slug>:<agency_type>, with the street part added only where one county has several offices of one kind or no county is known. Probation blocks that do not say adult or juvenile are typed probation_adult (the schema has no combined type) and named "Probation Office".

• https://ohflac.wvdhhr.org/Apps/Lookup/FacilitySearch/w1
• https://ohflac.wvdhhr.org/Apps/Lookup/FacilitySearch/FacilitySearch
• https://ohflac.wvdhhr.org/Apps/Lookup/FacilityDetails/<facility id> (one per kept facility)
• https://ohflac.wvdhhr.org/index.html

HTML + JSON over plain HTTP via lib/http.ts. The lookup page is a DataTables app whose own script POSTs a JSON filter to /Apps/Lookup/FacilitySearch/FacilitySearch (server-side paging; length -1 returns all rows) and links each row to /Apps/Lookup/FacilityDetails/<id>. List rows carry phone/fax/address/status/administrator; detail pages are fetched once per kept facility and cross-checked. robots.txt on ohflac.wvdhhr.org returns 404 (no restrictions). Rights pages checked: lookup footer (Privacy/Security/Accessibility + Copyright 2013 State of West Virginia), ohflac.wvdhhr.org/index.html, www.wv.gov/policies, /policies/disclaimer, /policies/legal-notices, /privacy-policy; none bars reuse, scraping or commercial use; wvdhhr.org root did not respond. Service-location facilities show no license of their own and point at the licensed parent facility.

• https://dcr.wv.gov/facilities/Pages/parole-services-offices.aspx
• https://dcr.wv.gov/facilities/Pages/parole-services-offices/<13 office pages>.aspx
• https://www.wv.gov/media/20/download?inline=

Rights: the wv.gov content disclaimer (https://www.wv.gov/policies/disclaimer) only disclaims warranty of accuracy and timeliness and states reliance is at the user's own risk; it states no reuse, copying or scraping terms, so no rights clause is recorded. HTML: the index page lists the parole offices; each office page was fetched via lib/http.ts (robots.txt honoured). PDF: WV State Government Directory (wv.gov/media/20/download?inline=), text extracted with lib/pdf.ts; the DCR central and Board of Parole entries were located by their exact directory headings and the single printed "Fax" line and the labelled "Phone" line of each read. Rights pages checked: dcr.wv.gov footer (Contact Us / Site Map / Staff Directory only; no site-level terms page of its own), dcr.wv.gov/aboutus/Pages/Policies.aspx (DCR operational policies, not site terms), wv.gov/policies (index), wv.gov/policies/disclaimer, /policies/legal-notices, /policies/link-policy, /privacy-policy. None bars reuse, commercial use or scraping; the disclaimer is an accuracy/liability disclaimer and pages carry a generic "(c) State of West Virginia" footer, so status stays public_unresolved. Mapping: county is the "(X County)" parenthetical printed under each office address (normalizeCounty); office_name is the page title as published; the Moorefield page lives at a Keyser-Parole-Office.aspx URL (kept as published).

• https://bbh.wv.gov/comprehensive-behavioral-health-centers
• https://www.wv.gov/policies/disclaimer
• https://www.wv.gov/policies
• https://www.wv.gov/policies/legal-notices
• https://www.wv.gov/privacy-policy

One static Drupal HTML page (bbh.wv.gov) fetched through lib/http.ts: the centers are a list of one-link paragraphs (link text = center name, href = center website). Identity rows only (office_name, website); county/address/department null because the page prints none that are the center's own (the footer address is the Bureau's own office); no routes. Websites are stored exactly as published, including 3 that are not the center's own domain (HealthWays -> rehab.com listing, United Summit Center -> wvumedicine.org page, Mountain Laurel -> trailing "#"). Rights pages checked: bbh.wv.gov page footer (only link is Privacy, Security and Accessibility -> wv.gov/policies), wv.gov/policies, /policies/disclaimer, /policies/legal-notices, /policies/link-policy, /privacy-policy. None bars reuse, commercial use or scraping; the disclaimer only disclaims accuracy/timeliness and states reliance is at the user's own risk (it states no reuse terms, so no rights clause is recorded); no reuse licence is granted either, hence public_unresolved. parent_org_rules name patterns were derived from the BBH link text plus the centers' own sites (about/history pages: FMRS former name, Mountain Laurel former name, Westbrook former name, SHCMHC short name); prestera.org returned HTTP 403 and wvumedicine.org is a blocked host, so those two rest on the BBH link text only.

• https://www.camc.org/patients-and-visitors/patient-information/medical-records
• https://www.camc.org/sites/default/files/MedicalRecords/17-8445%20FINAL%20ROI%20Form%203%20page%20Minor%20Law%20changes-7.11.2025.pdf
• https://www.camc.org/privacy-policy
• https://chh.marshallhealthnetwork.org/customer-service/medical-records
• https://chh.marshallhealthnetwork.org/terms-conditions
• https://cabincreekhealth.com/medical-records-transfer/
• https://valleyhealth.org/patient-info/medical-record-release-forms/
• https://www.swellbox.com/valley-health-systems-provider-wizard.html
• https://wvumedicine.org/patient-resources/records-and-planning/request-medical-records/

HAND-ENTERED constants taken from four public HTML pages (CAMC, Cabell Huntington Hospital, Cabin Creek Health Systems, Valley Health) plus the CAMC "Authorization of Use and Disclosure" PDF form that the CAMC page links (it holds the authorization fax numbers; the CAMC page itself says to fax "to the applicable fax number listed on the form"). Each run re-fetches the pages through lib/http.ts (cache) and checks every hand-entered number/string against the fetched text; misses are logged as DRIFT failures. Valley Health provider portal URL is stored exactly as the ticket gives it; it answers with a redirect which is read with redirect:"manual" and recorded in notes only (not followed). WVU Medicine (wvumedicine.org) is restricted_no_scrape: ONE office row, no routes, host never fetched. Rights pages checked: CHH https://chh.marshallhealthnetwork.org/terms-conditions (content informational only, no reuse/scraping bar); CAMC https://www.camc.org/privacy-policy (HIPAA notice only, no site terms); site footers of camc.org, cabincreekhealth.com (All Rights Reserved) and valleyhealth.org (Copyright by ValleyHealth.org); the conventional /terms-of-use, /terms-and-conditions and /privacy-policy paths on www.camc.org, cabincreekhealth.com and valleyhealth.org are probed on every run (those that fail are listed under Failed fetches in import-report.md; none gave a terms page).

• https://dpscs.maryland.gov/locations/dpp_offices.shtml

robots.txt of dpscs.maryland.gov read before fetching (no path used here is disallowed). The page footer links only privacy, accessibility and contact pages; none states a reuse term, so no rights clause is recorded and the status stays public_unresolved. One HTTP GET of the office table page through lib/http.ts. Each table row is one office (county headings are the rows with a single colspan cell); the headquarters is the bold paragraph under the table. Mapping: county = the county heading above the row (Baltimore City and Baltimore County are separate); office_name as published without the trailing asterisk; address = the address cell lines joined with commas, "Maryland" and "MD." written MD; every phone and every fax is its own route; a number with a parenthesised label goes on a sibling office row (same name and address) whose department is the label; "N/A" yields no route; the Directions column is not read.

• https://dhs.maryland.gov/local-offices/
• https://dhs.maryland.gov/local-offices/<24 county and Baltimore City pages>/
• https://www.maryland.gov/pages/privacy_security.aspx

robots.txt of dhs.maryland.gov read before fetching (no path used here is disallowed). Terms: the pages link only the State of Maryland privacy and accessibility pages (maryland.gov/pages/privacy_security.aspx, read: a personal-information notice, no reuse, copying or scraping term), so no rights clause is recorded and the status stays public_unresolved. The 24 page URLs are read from the /local-offices/ index. Per page the DSS block is the section under the first heading naming the DSS (or "Administration Building"), the child support block the section under "Child Support Administration", and the service table the table under "Contact Information". DSS: address (leading building names dropped; where a page prints mailing and physical addresses the physical one), the number under Phone / Main Phone / Main Line, every number under Fax / Facsimile (a fax with sub-labels becomes one row per label, the label being the department), hours of Operation or Lobby Hours as published. Child support: only a number printed under Local / Main Phone / Child Support Unit Phone becomes a route; the 1-800 customer service number, the 1-844 payment line and TTY numbers are not routes. Service table: a row's number is the one displayed in its Phone cell; rows with the same number form one office row (same name and address as the DSS) with department "S1, S2, ..."; the statewide 1-800-917-7383 rows form one row per office with department "Reporting suspected abuse or neglect". Emails (obfuscated by the site and never decoded), staff and official names, photo captions and links are not read into any field.

• https://msa.maryland.gov/msa/mdmanual/32dc/html/dccounty.html
• https://msa.maryland.gov/msa/mdmanual/36loc/<24 county judicial pages>/

robots.txt of msa.maryland.gov read before fetching (no path used here is disallowed). The page footer states the Manual is presented for reference under fair use and that citation and credit to the Maryland State Archives are required; that credit is shown on /sources/ and beside every route taken from this source. The 24 county page URLs are read from the District Court "Judges by County" index. Per page: circuit clerk office = the CLERK block (Baltimore City: CLERK'S OFFICE), phone and fax from the office-level number line (TDD/TTY numbers are not routes); toll-free numbers are kept as phone routes; a department line (civil, criminal, juvenile, land records) becomes a sibling office row with that department only where no person is attached to it; District Court = the number(s) printed on the County Clerk line; orphans' court = the address and number line under its heading; register of wills = the number line under its heading. Sheriffs, detention, court administration, magistrates, judges, staff, every name and every email address are not read into any field (names go to persons_dropped).

• https://dam.assets.ohio.gov/image/upload/jfs.ohio.gov/County/County_Directory.pdf

One PDF, the County Agency Directory of the Ohio Department of Job and Family Services (ODJFS), "Updated: February 6, 2023" on page 1 and "Rev. 2/6/2023" in the footer of every county page: the state's current edition. Fetched through lib/http.ts from dam.assets.ohio.gov (robots.txt honoured by the fetcher; the PDF path is not disallowed). Terms: neither the PDF nor its pages state a reuse, copying or scraping term, so no rights clause is recorded and the status stays public_unresolved. jfs.ohio.gov/robots.txt answered HTTP 404 from the build machine on 2026-10-23; the text recorded for it is the one reported from another network: it disallows only /wps/wcm/connect/gov/ohio+content+english/ (listed under two spellings of that path). No page of jfs.ohio.gov is fetched here. emanuals.jfs.ohio.gov (the ODJFS eManuals county directory) is unreachable from the build machine (the TLS connection is reset on every attempt, last tried 2026-10-23) and is not used. Text extracted with lib/pdf.ts (one page per county, "<County> County - NN"). Per agency block: heading "CDJFS|CSEA|PCSA - <title>" (a heading that starts with "- " and no acronym takes the type of the block above it, except "Adult Protective Services", which the directory's own introduction assigns to the CDJFS), a director line (dropped), the address, then the labelled fields Phone/Ext, Fax, Hours, Website; a field may wrap onto the next line. A Fax field may carry "TTY/TTD:" and "After Hours Emergency:" parts; the same split is applied to the Phone/Ext field. Numbers: slash shorthand is expanded ("(513) 887-4296/4334" is two faxes, "(740) 947-5080/7761/7762" three phones: the last four digits are replaced); a vanity number takes the digits in its parentheses ("296-CARE(2273)" under (330) is 330-296-2273, "64-CHILD (642-4453)" under (937) is 937-642-4453); an extension goes into the phone value ("ext"), a fax never carries one; toll-free numbers (800, 833, 844, 855, 866, 877, 888) are phone routes with conditions "Toll-free"; "Option 3" is copied to conditions. A word label before a number ("Adm", "Collabor8", "Child Care Network") is the department of a sibling office row; TTY/TTD numbers are phone routes on a sibling row with department "TTY/TTD"; After Hours Emergency numbers are phone routes on a sibling row with department "After-hours emergency" (with the printed label added in parentheses, for example Sheriff). Hours are copied as printed onto the main phone and fax routes of the block. Websites without a scheme are given https:// (bare domains). Not read into any field: directors and administrators, the OhioMeansJobs (OMJ) blocks, the adoption lines (the state adoption number and each county adoption contact, whose names are dropped), 911, and TTY values that are not numbers ("Yes", "Available on request", 7-1-1).