Mirri additions

CHANGELOG.md

@@ -1,3 +1,39 @@
+## v0.10.0 (2026-03-12)
+
+### Feat
+
+- **acr_db.json,regex_db.json**: complete list of CBS ccno examples for all found suffixes
+- **acr_db.json,regex_db.json**: add entry for ICCF Collection of Industrial Microorganisms
+- **acr_db.json,regex_db.json**: add entry for MSCL Microbial Strain Collection of Latvia
+- **acr_db.json,regex_db.json**: add entry for UNIMORE Microbial Culture Collection (UMCC)
+- **acr_db.json,regex_db.json**: add entry for PYCC The portuguese yeast culture collection
+- **acr_db.json,regex_db.json**: add entry for UCCCB The University of Coimbra Bacteria Culture Collection
+- **acr_db.json,regex_db.json**: add entry for UOA/HCPF Hellenic Collection of Pathogenic Fungi
+- **acr_db.json,regex_db.json**: add entry for ACOI Coimbra Collection of Algae
+- **acr_db.json,regex_db.json**: add entry for The Athens University Bacterial & Archaea Culture Collection ATHUBA
+- **acr_db.json,regex_db.json**: add entry for ITEM Agri-Food Microbial Collection
+- add MIRRI acronym
+
+### Fix
+
+- **acr_db.json**: fixed ACOI suffix regex correctly
+- **acr_db.json**: changed ACOI suffix regex from [] to (|)
+- **acr_db.json,catalogue_db.json,regex_db.json**: fixed mismatching and incomplete regexes for ACOI
+- **acr_db.json**: fix mismatching regexes for CCUG
+- **acr_db.json**: fixed mismatching regexes for UOA/HCPF
+- **acr_db.json,catalogue_db.json,regex_db.json**: fixed some issues with regexes and removed MIRRI acronym
+- **acr_db.json,regex_db.json**: recognize CBS ccnos with capital letter suffix
+- **catalogue_db.json**: add missing ccnos for new entries with catalogue urls
+- **acr_db.json,regex_db.json**: recognize LEGE ccnos with one dot
+- **acr_db.json,regex_db.json**: recognize suffix 'i' for MUM collection
+- **acr_db.json,regex_db.json**: recognize BEA ccnos with one slash
+- **regex_db.json**: add private CCUG ccno examples
+- **acr_db.json**: Extended CCUG regex to accept ! and #
+
+### Refactor
+
+- **validate**: adjust regex patterns and error message
+
 ## v0.9.15 (2026-01-12)
 
 ### Refactor

README.md

@@ -1,6 +1,6 @@
 # Collection acronyms for identification - cafi
 
-[![release: 0.9.15](https://img.shields.io/badge/rel-0.9.15-blue.svg?style=flat-square)](https://github.com/LeibnizDSMZ/cafi)
+[![release: 0.10.0](https://img.shields.io/badge/rel-0.10.0-blue.svg?style=flat-square)](https://github.com/LeibnizDSMZ/cafi)
 [![MIT LICENSE](https://img.shields.io/badge/License-MIT-brightgreen.svg?style=flat-square)](https://choosealicense.com/licenses/mit/)
 [![DATA LICENSE - CC BY 4.0](https://img.shields.io/badge/Data%20License-CC%20BY%204.0-brightgreen.svg?style=flat-square)](http://creativecommons.org/licenses/by/4.0/)
 [![Documentation Status](https://img.shields.io/badge/docs-GitHub-blue.svg?style=flat-square)](https://LeibnizDSMZ.github.io/cafi/)

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "cafi"
-version = "0.9.15"
+version = "0.10.0"
 description = "This project acts as a registry for acronyms used by culture collections."
 readme = "README.md"
 authors = [{ name = "Artur Lissin", email = "artur.lissin@dsmz.de" }]

src/cafi/constants/versions.py

@@ -1,4 +1,4 @@
 from typing import Final
 
-CURRENT_VER: Final[str] = "v0.9.15"
+CURRENT_VER: Final[str] = "v0.10.0"
 STABLE_VER: Final[str] = "main"

src/cafi/data/acr_db.json

@@ -366,10 +366,10 @@
     "country": "PT",
     "homepage": "https://lege.ciimar.up.pt",
     "name": "Blue Biotechnology and Ecotoxicology Culture Collection",
-    "regex_ccno": "^LEGE\\s*(XX?-?)?\\d+M?$",
+    "regex_ccno": "^LEGE\\s*(XX?-?)?\\d+(\\.\\d+)?M?$",
     "regex_id": {
-      "core": "\\d+",
-      "full": "^(XX?-?)?\\d+M?$",
+      "core": "\\d+(\\.\\d+)?",
+      "full": "^(XX?-?)?\\d+(\\.\\d+)?M?$",
       "pre": "XX?",
       "suf": "M"
     },
@@ -486,10 +486,10 @@
     "country": "ES",
     "homepage": "https://marinebiotechnology.org/en/beaen/about-us.html",
     "name": "Spanish Bank of Algae",
-    "regex_ccno": "^BEA\\s*\\d+B?$",
+    "regex_ccno": "^BEA\\s*\\d+(/\\d+)?B?$",
     "regex_id": {
-      "core": "\\d+",
-      "full": "^\\d+B?$",
+      "core": "\\d+(/\\d+)?",
+      "full": "^\\d+(/\\d+)?B?$",
       "suf": "B"
     },
     "ror": "01teme464"
@@ -639,6 +639,144 @@
     },
     "ror": "02y5sbr94"
   },
+  "130": {
+    "acr": "ITEM",
+    "acr_synonym": [
+      "ISPA:ITEM",
+      "CNR:ISPA:ITEM"
+    ],
+    "active": true,
+    "code": "CNR:ISPA:ITEM",
+    "country": "IT",
+    "gbif": "21c9b326-b756-4391-b302-f3e095b947cb",
+    "homepage": "https://item.bio-aware.com/",
+    "name": "ITEM Agri-Food Microbial Collection",
+    "regex_ccno": "^ITEM\\s*\\d+$",
+    "regex_id": {
+      "full": "^\\d+$"
+    }
+  },
+  "131": {
+    "acr": "ATHUBA",
+    "active": true,
+    "code": "ATHUBA",
+    "country": "GR",
+    "gbif": "1a74161f-2a7b-4f40-8e4a-c7d2db88ae2c",
+    "homepage": "http://m-biotech.biol.uoa.gr/ATHUBstrains.html",
+    "name": "The Athens University Bacterial & Archaea Culture Collection",
+    "regex_ccno": "^ATHUBA\\s*\\d+$",
+    "regex_id": {
+      "full": "^\\d+$"
+    }
+  },
+  "132": {
+    "acr": "ACOI",
+    "active": true,
+    "catalogue": [
+      "http://acoi.ci.uc.pt/spec_detail.php?cult_id={id}"
+    ],
+    "code": "ACOI",
+    "country": "PT",
+    "gbif": "e3fba94d-bd32-4a35-9385-a97263f618b1",
+    "homepage": "http://acoi.ci.uc.pt/index.php",
+    "name": "ACOI Coimbra Collection of Algae",
+    "regex_ccno": "^ACOI\\s*(-\\s*)?\\d+(-(A|B))?$",
+    "regex_id": {
+      "core": "\\d+",
+      "full": "^\\d+(-(A|B))?$",
+      "suf": "(A|B)"
+    }
+  },
+  "133": {
+    "acr": "UOA:HCPF",
+    "active": true,
+    "code": "UOA:HCPF",
+    "country": "GR",
+    "gbif": "e9e55bd6-98d7-41a7-a250-da49ded6dc94",
+    "homepage": "http://www.med.uoa.gr/~aveleg/index_files/Page596.htm",
+    "name": "UOA/HCPF Hellenic Collection of Pathogenic Fungi",
+    "regex_ccno": "^UOA/HCPF\\s*(REF|ENV|AM)?\\s*\\d+([.-]\\d+)*[\\s-]*(A|B|C|a|b)?$",
+    "regex_id": {
+      "core": "\\d+([.-]\\d+)*",
+      "full": "^(REF|ENV|AM)?\\s*\\d+([.-]\\d+)*[\\s-]*(A|B|C|a|b)?$",
+      "pre": "(REF|ENV|AM)",
+      "suf": "(A|B|C|a|b)"
+    }
+  },
+  "134": {
+    "acr": "UCCCB",
+    "active": true,
+    "catalogue": [
+      "https://ucccb.uc.pt/strain-details/?detail={acr}{id}"
+    ],
+    "code": "UCCCB",
+    "country": "PT",
+    "homepage": "https://ucccb.uc.pt/",
+    "name": "UCCCB The University of Coimbra Bacteria Culture Collection",
+    "regex_ccno": "^UCCCB\\s*\\d+$",
+    "regex_id": {
+      "full": "^\\d+$"
+    }
+  },
+  "135": {
+    "acr": "PYCC",
+    "acr_synonym": [
+      "NOVA:PYCC"
+    ],
+    "active": true,
+    "catalogue": [
+      "https://pycc.pt/pycc-{id}"
+    ],
+    "code": "PYCC",
+    "country": "PT",
+    "homepage": "https://pycc.pt/",
+    "name": "PYCC The portuguese yeast culture collection",
+    "regex_ccno": "^PYCC\\s*\\d+$",
+    "regex_id": {
+      "full": "^\\d+$"
+    }
+  },
+  "136": {
+    "acr": "UMCC",
+    "active": true,
+    "code": "UMCC",
+    "country": "IT",
+    "homepage": "https://umcc.bio-aware.com/page/Homepage",
+    "name": "UNIMORE Microbial Culture Collection (UMCC)",
+    "regex_ccno": "^UMCC\\s*\\d+$",
+    "regex_id": {
+      "full": "^\\d+$"
+    }
+  },
+  "137": {
+    "acr": "MSCL",
+    "active": true,
+    "code": "MSCL",
+    "country": "LV",
+    "gbif": "26bed235-564f-4c34-a971-dafcc7371b2a",
+    "homepage": "https://www.lu.lv/en/mbi/laboratories/microbial-strain-collection-of-latvia",
+    "name": "MSCL Microbial Strain Collection of Latvia",
+    "regex_ccno": "^MSCL\\s*\\d+$",
+    "regex_id": {
+      "full": "^\\d+$"
+    }
+  },
+  "138": {
+    "acr": "ICCF",
+    "acr_synonym": [
+      "CMII:ICCF",
+      "CMII:ICCF:WFCC"
+    ],
+    "active": false,
+    "code": "ICCF",
+    "country": "RO",
+    "gbif": "12693a23-a500-41c8-a7c6-61403ea24900",
+    "name": "ICCF Collection of Industrial Microorganisms",
+    "regex_ccno": "^ICCF\\s*\\d+$",
+    "regex_id": {
+      "full": "^\\d+$"
+    }
+  },
   "14": {
     "acr": "JCM",
     "active": true,
@@ -954,11 +1092,11 @@
     "gbif": "323cc3eb-16f5-4aa0-8126-f0ccdfe9b56d",
     "homepage": "https://www.ccug.se/",
     "name": "Culture Collection University of Gothenburg",
-    "regex_ccno": "^CCUG\\s*\\d+(\\s*[AaBbCEQDFG]{1,2})?$",
+    "regex_ccno": "^CCUG\\s*\\d+(\\s*[AaBbCEQDFG!#]{1,2})?$",
     "regex_id": {
       "core": "\\d+",
-      "full": "^\\d+(\\s*[AaBbCEQDFG]{1,2})?$",
-      "suf": "[AaBbCEQDFG]{1,2}"
+      "full": "^\\d+(\\s*[AaBbCEQDFG!#]{1,2})?$",
+      "suf": "[AaBbCEQDFG!#]{1,2}"
     },
     "ror": "01tm6cn81"
   },
@@ -1359,9 +1497,11 @@
     "gbif": "1370bc72-540f-4ff0-aefd-2358971299be",
     "homepage": "https://wi.knaw.nl/",
     "name": "Westerdijk Institute: CBS Collection of yeasts, filamentous fungi and bacteria",
-    "regex_ccno": "^CBS\\s*\\d+(\\.\\d+)?$",
+    "regex_ccno": "^CBS\\s*\\d+(\\.\\d+)?[A-Z]?$",
     "regex_id": {
-      "full": "^\\d+(\\.\\d+)?$"
+      "core": "\\d+(\\.\\d+)?",
+      "full": "^\\d+(\\.\\d+)?[A-Z]?$",
+      "suf": "[A-Z]"
     },
     "ror": "030a5r161"
   },
@@ -1666,9 +1806,11 @@
     "gbif": "dfdd9fce-e4b9-4444-a86b-bb68b2b2ef48",
     "homepage": "https://www.micoteca.deb.uminho.pt",
     "name": "Micoteca da Universidade do Minho",
-    "regex_ccno": "^MUM\\s*\\d+(\\.\\d+)?$",
+    "regex_ccno": "^MUM\\s*\\d+(\\.\\d+)?i?$",
     "regex_id": {
-      "full": "^\\d+(\\.\\d+)?$"
+      "core": "\\d+(\\.\\d+)?",
+      "full": "^\\d+(\\.\\d+)?i?$",
+      "suf": "i"
     }
   },
   "67": {

src/cafi/data/catalogue_db.json

@@ -65,6 +65,17 @@
   "129": [
     "UIO 457"
   ],
+  "132": [
+    "ACOI 2754",
+    "ACOI 2902-A",
+    "ACOI 2902-B"
+  ],
+  "134": [
+    "UCCCB 231"
+  ],
+  "135": [
+    "PYCC 10001"
+  ],
   "14": [
     "JCM 12345"
   ],
@@ -148,7 +159,27 @@
     "NCIM 1001"
   ],
   "5": [
-    "CBS 101"
+    "CBS 3",
+    "CBS 101",
+    "CBS 107.95",
+    "CBS 208.77A",
+    "CBS 528.79B",
+    "CBS 721.73C",
+    "CBS 287.70D",
+    "CBS 226.71E",
+    "CBS 377.70F",
+    "CBS 907.72G",
+    "CBS 378.70H",
+    "CBS 907.72I",
+    "CBS 298.70J",
+    "CBS 287.70K",
+    "CBS 295.70L",
+    "CBS 376.70M",
+    "CBS 295.70N",
+    "CBS 287.70O",
+    "CBS 377.70P",
+    "CBS 377.70Q",
+    "CBS 287.70R"
   ],
   "50": [
     "RCC103"