String & encoding developer: MQL recipes

Recipes

Clean a customer line before import
SQL table and API names from form labels
Parse an API access log line
Encode a payment callback and check it
Colors, permissions and character codes
Encrypt, decrypt and sign a secret
Match a typed product name to the catalog
Mail merge: fill a template from JSON data
Mask personal data before writing a log line
Write and read a fixed width bank transfer file

executed on MentDB

Clean a customer line before import

A customer file arrives with extra spaces, mixed case, accents and an empty note. The script trims and normalises each field, builds the initials, a padded customer code and a slug, and outputs a clean CSV line.

Pipeline

01 · trimTrim and collapse spacesstring itrim→02 · caseCapitalise the namestring capitalize_words→03 · mailLower case, no accentstring unaccent→04 · codePadded customer codestring lpad→05 · csvCSV valuesstring csv_value

Script

admin

#1. A raw line of a customer file with spaces, accents and mixed case;
-> "[name]" "  jean-PAUL    dupont-lefèvre ";
-> "[mail]" " ÉLODIE.Martin@Mail.COM  ";
-> "[city]" "   Saint-Denis   ";
-> "[country]" " FRANCE ";
-> "[code]" "  042  ";
-> "[note]" "";
#2. Clean every field;
-> "[name]" (string capitalize_words (string itrim (string lrtrim [name])));
-> "[mail]" (string lcase (string unaccent (string trim [mail])));
-> "[city]" (string upper (string rtrim (string ltrim [city])));
-> "[code]" (concat "C" (string lpad (string lrtrim [code]) "0" 6));
-> "[initials]" (concat (string ucase (string first_letter [name])) (string upper (string first_letter (string mid [name] (+ (string instr [name] " ") 1)))));
#3. Optional field, slug and CSV export;
-> "[note]" (string empty_if_null (string null_if_empty [note]));
concat (string csv_value [code] ";" "\"") ";" (string csv_value [name] ";" "\"") ";" [initials] ";" [mail] ";" [city] ";" (string first_letter_upper (string lower (string trim [country]))) ";" (string slugify [name]) ";" (string zero [note]) ";blank=" (string is_blank [note]) ";" (string lrtrim0 "00042")

Result

mentdb

C000042;Jean-Paul Dupont-Lefèvre;JD;elodie.martin@mail.com;SAINT-DENIS;France;jean-paul-dupont-lefevre;0;blank=1;42

Go further

Detect duplicates with string levenshtein_distance on the cleaned names
Check the cleaned mail with type is_email before the insert

executed on MentDB

SQL table and API names from form labels

A form has French labels with accents and spaces. The script turns them into snake_case column names aligned with padding, SQL comments with escaped quotes, a camelCase JSON field, a kebab-case route and an HTML view of the MQL snippet for the wiki.

Pipeline

01 · namesnake_case columnsstring snake_case→02 · alignAlign with paddingstring rpad→03 · commentSQL literalstring txt2→04 · apicamelCase and kebab-casestring camel_case→05 · wikiMQL to HTMLstring mql_to_html

Script

admin

#1. Business labels of a form become a SQL table;
-> "[labels]" "Date de naissance,Montant TTC,Numéro client,Commentaire libre du vendeur";
-> "[ddl]" (concat "CREATE TABLE " (string snake_case "Fiche Client") " (" [_n_]);
-> "[i]" 0;
#2. One column per label: snake_case name, padded for alignment, SQL comment;
for (-> "[i]" 1) (<= [i] 4) (++ "[i]") {
  -> "[lab]" (atom get [labels] [i] ",");
  -> "[col]" (string snake_case (string unaccent [lab]));
  -> "[ddl]" (concat [ddl] (string rpad [col] " " 30) "VARCHAR(100) COMMENT " (string txt2 (string truncate [lab] 20)) "," [_n_]);
};
-> "[ddl]" (concat (string substring [ddl] 0 (- (string length [ddl]) 2)) [_n_] ");");
#3. Same names for the JSON API and the URL routes, plus the snippet in HTML for the wiki;
concat (string indent [ddl] 2) [_n_] (string repeat "-" 40) [_n_] "json: " (string camel_case (string unaccent "Montant TTC")) ", route: /" (string kebab_case (string camel_case (string unaccent "Numéro client"))) [_n_] "sql literals: " (string to_string "O\"Neil") " " (string txt "l'été") [_n_] "trimmed: " (string sublrchar "[ok]" 1) (string space 2) (string replace "v1.2.3" "." "_") [_n_] (string del_char_before_each_line (concat "-- col1" [_n_] "-- col2") 3) [_n_] "doc html: " (string mql_to_html (mql {string snake_case "Fiche Client";}))

Result

mentdb

CREATE TABLE fiche_client (
  date_de_naissance             VARCHAR(100) COMMENT 'Date de naissance',
  montant_ttc                   VARCHAR(100) COMMENT 'Montant TTC',
  numero_client                 VARCHAR(100) COMMENT 'Numéro client',
  commentaire_libre_du_vendeur  VARCHAR(100) COMMENT 'Commentaire libre...'
  );
----------------------------------------
json: montantTtc, route: /numero-client
sql literals: 'O"Neil' 'l\'été'
trimmed: ok  v1_2_3
col1
col2

doc html: <span style='color:#00baff'>string</span> <span style='color:#00baff'>snake_case</span> <span style='color:#fff'>"Fiche Client"</span>;

Go further

Pick the column types with the type is_* checks on sample data
string truncate keeps comments under the database limit

executed on MentDB

Parse an API access log line

An access log line holds a date, a level, a route with an order id, a status and a duration. The script extracts the fields with regex groups, rewrites the date, reads the variables of an error message with a pattern and classifies the line with the match functions.

Pipeline

01 · extractRegex groupsstring regexp_extract→02 · rewriteDate to dd/MM/yyyystring regexp_replace→03 · patternVariables of a messagestring get_variable→04 · splitWords and partsstring split→05 · classifyMatch and containsstring regexp

Script

admin

#1. One line of an API access log;
-> "[l]" "2026-11-01 08:15:42 ERROR [api] GET /orders/A-1234?user=jim 500 812ms";
#2. Extract the fields with regex groups;
-> "[code]" (string regexp_extract [l] " ([0-9]{3}) [0-9]+ms$" 1);
-> "[ms]" (string regexp_extract [l] "([0-9]+)ms$" 1);
-> "[order]" (string regexp_extract [l] "/orders/([A-Z]-[0-9]+)" 1);
-> "[day]" (string regexp_replace (string left [l] 10) "([0-9]+)-([0-9]+)-([0-9]+)" "$3/$2/$1");
-> "[msg]" "Timeout in module payment at line 42";
-> "[vars]" (string replace (string replace (string get_variable [msg] "[1] in module [2] at line [3]") [_n_] "") "  " "");
#3. Classify the line;
-> "[words]" (string split [l] " " -1);
concat "{\"day\": \"" [day] "\", \"order\": \"" [order] "\", \"status\": " [code] ", \"ms\": " [ms] ", \"message_vars\": " [vars] ", \"words\": " [words] ", \"numbers\": " (string regexp_count [l] "[0-9]+") ", \"slashes\": " (string count [l] "/") ", \"is_error\": " (string contains [l] "ERROR") ", \"has_user\": " (string contains_ignore_case [l] "USER=") ", \"api_tag\": " (string regexp [l] ".*\[api\].*") ", \"not_debug\": " (string not_regexp [l] ".*DEBUG.*") ", \"like_get\": " (string like [l] ".* GET .*") ", \"not_like_post\": " (string not_like [l] ".* POST .*") ", \"matches\": " (string matches [l] ".*orders.*") ", \"is_matches_regex\": " (type is_matches_regex [l] ".*812ms") ", \"starts_2026\": " (string starts_with [l] "2026") ", \"starts_digit\": " (string starts_with_or [l] "0123456789") ", \"ends_ms\": " (string ends_with [l] "ms") ", \"level_is_error\": " (string equals_ignore_case (string split_mql [l] 2) "error") ", \"parts\": " (string split_sentence "GET /orders/A-1234?user=jim" "/?=") "}"

Result

mentdb

{"day": "01/11/2026", "order": "A-1234", "status": 500, "ms": 812, "message_vars": ["Timeout","payment","42"], "words": ["2026-11-01","08:15:42","ERROR","[api]","GET","\/orders\/A-1234?user=jim","500","812ms"], "numbers": 9, "slashes": 2, "is_error": 1, "has_user": 1, "api_tag": 1, "not_debug": 1, "like_get": 1, "not_like_post": 1, "matches": 1, "is_matches_regex": 1, "starts_2026": 1, "starts_digit": 1, "ends_ms": 1, "level_is_error": 1, "parts": ["GET","orders","A-1234","user","jim"]}

Go further

Run it on each line read with file reader_get_line and count the 500 errors per hour
string regexp_count gives a quick density of numbers or ids per line

executed on MentDB

Encode a payment callback and check it

A payment callback carries a French label with & and accents, a Basic auth header and an HTML note. The script URL-encodes and Base64-encodes them, escapes for HTML and JSON, decodes everything back and measures the label.

Pipeline

01 · urlURL encode the labelstring url_encode→02 · authBasic auth in Base64string encode_b64→03 · htmlEscape for HTMLstring html_escape→04 · jsonEscape for JSONstring json_escape→05 · backDecode backstring url_decode

Script

admin

#1. A payment callback with a French label and an HTML note;
-> "[label]" "Commande n°42 & frais d'été";
-> "[q]" (concat "label=" (string url_encode [label]) "&amount=19.90");
-> "[auth]" (concat "Basic " (string encode_b64 "shop:test-key"));
#2. Safe versions for HTML and JSON outputs;
-> "[html]" (string html_escape "<b>Tom & Jerry's</b>");
-> "[js]" (string json_escape (concat "line1" [_n_] "say \"hi\""));
#3. Decode everything back and measure the label;
concat "{\"query\": \"" [q] "\", \"auth\": \"" [auth] "\", \"auth_back\": \"" (string decode_b64 (string mid [auth] 6)) "\", \"label_back\": \"" (string url_decode (string mid [q] 6 (string instr [q] "&a"))) "\", \"html\": \"" [html] "\", \"html_back\": \"" (string html_unescape "caf&#233; &amp; th&#xE9;") "\", \"json_value\": \"" [js] "\", \"utf8_to_latin1\": \"" (string encode "abc" "UTF-8" "ISO-8859-1") "\", \"length\": " (string length [label]) ", \"char_length\": " (string char_length [label]) ", \"bit_length\": " (string bit_length [label]) "}"

Result

mentdb

{"query": "label=Commande%20n%C2%B042%20%26%20frais%20d%27%C3%A9t%C3%A9&amount=19.90", "auth": "Basic c2hvcDp0ZXN0LWtleQ==", "auth_back": "shop:test-key", "label_back": "Commande n°42 & frais d'été", "html": "&lt;b&gt;Tom &amp; Jerry&#39;s&lt;/b&gt;", "html_back": "café & thé", "json_value": "line1\nsay \"hi\"", "utf8_to_latin1": "abc", "length": 27, "char_length": 27, "bit_length": 216}

Go further

Sign the query with string hmac_sha256 so the shop can check it
string bit_length tells the size in bits of the payload, string length the number of characters

executed on MentDB

Colors, permissions and character codes

A front developer needs the RGB values of the brand color #FFB547 and a sysadmin wants to read the 755 permission in decimal and in bits. The script converts between hexadecimal, octal, binary and characters.

Pipeline

01 · hexHex pairs to integersstring hex_to_int→02 · backBack to hexstring int_to_hex→03 · octalOctal permissionstring oct_to_int→04 · bitsBinary formstring bin→05 · charsCharacters and codesstring char_to_int

Script

admin

#1. A web color and a Unix permission to decode;
-> "[color]" "FFB547";
-> "[r]" (string hex_to_int (string left [color] 2));
-> "[g]" (string unhex (string mid [color] 2 4));
-> "[b]" (string hex_to_int (string right [color] 2));
-> "[perm]" (string oct_to_int "755");
#2. Characters and their codes;
-> "[c]" "é";
concat "{\"rgb\": [" [r] ", " [g] ", " [b] "], \"back_to_hex\": \"" (string upper (concat (string hex [r]) (string int_to_hex [g]) (string int_to_hex [b]))) "\", \"perm_755_decimal\": " [perm] ", \"perm_bits\": \"" (string bin [perm]) "\", \"perm_octal\": \"" (string oct [perm]) "|" (string int_to_oct 420) "\", \"ascii_A\": " (string ascii "A") ", \"code_e_acute\": " (string char_to_int [c]) ", \"char_77\": \"" (string char 77) (string int_to_char 68) (string int_to_char 66) "\", \"hex_of_MDB\": \"" (string str_to_hex "MDB") "\", \"hex_41_to_str\": \"" (string hex_to_str "41") "\"}"

Result

mentdb

{"rgb": [255, 181, 71], "back_to_hex": "FFB547", "perm_755_decimal": 493, "perm_bits": "111101101", "perm_octal": "755|644", "ascii_A": 65, "code_e_acute": 233, "char_77": "MDB", "hex_of_MDB": "4d4442", "hex_41_to_str": "A"}

Go further

Lighten a color by adding 20 to each channel before string int_to_hex
string str_to_hex helps to spot invisible characters in an imported field

executed on MentDB

Encrypt, decrypt and sign a secret

An API token must be stored encrypted and its origin proven. The script generates RSA and signature key pairs, encrypts the token with RSA, DES, Blowfish and PBE, decrypts each one, signs it and checks that a tampered token fails the signature.

Pipeline

01 · keysKey pairsstring encode_rsa_generate_keypair→02 · rsaRSA encryptstring encode_rsa→03 · symDES, Blowfish, PBEstring encode_des→04 · signSignaturestring encode_sign→05 · verifyVerifystring decode_sign_verify

Script

admin

#1. Key pairs for RSA and for signatures, kept as JSON in memory;
json load "rx_rsa" (string encode_rsa_generate_keypair "2048");
json load "rx_sig" (string encode_sign_generate_keypair "2048");
-> "[secret]" (concat "token-" (string uuid));
#2. Encrypt with RSA, DES, Blowfish and PBE, then decrypt;
-> "[rsa]" (string encode_rsa [secret] (json select "rx_rsa" "/publicKey") (json select "rx_rsa" "/privateKey"));
-> "[des_key]" (string encode_des_generate_key "56");
-> "[des]" (string encode_des [secret] [des_key]);
-> "[bf]" (string encode_blowfish [secret] "vault-key-2026");
-> "[pbe]" (string encode_pbe [secret] "master-password");
#3. Sign the secret and check the signature;
-> "[sig]" (string encode_sign [secret] (json select "rx_sig" "/privateKey"));
-> "[ok]" (concat "{\"rsa_ok\": " (string equals_ignore_case (string decode_rsa [rsa] (json select "rx_rsa" "/privateKey")) [secret]) ", \"des_ok\": " (string equals_ignore_case (string decode_des [des] [des_key]) [secret]) ", \"blowfish_ok\": " (string equals_ignore_case (string decode_blowfish [bf] "vault-key-2026") [secret]) ", \"pbe_ok\": " (string equals_ignore_case (string decode_pbe [pbe] "master-password") [secret]) ", \"signature_ok\": " (string decode_sign_verify [secret] [sig] (json select "rx_sig" "/publicKey")) ", \"tampered_ok\": " (string decode_sign_verify (concat [secret] "x") [sig] (json select "rx_sig" "/publicKey")) ", \"random_salt_length\": " (string length (string generate_random_str 16)) "}");
json unload "rx_rsa";
json unload "rx_sig";
[ok]

Result

mentdb

{"rsa_ok": 1, "des_ok": 1, "blowfish_ok": 1, "pbe_ok": 1, "signature_ok": 1, "tampered_ok": 0, "random_salt_length": 16}

Go further

Keep the private keys out of the code: read them from a parameter
For passwords, prefer string bcrypt_hash, which cannot be decrypted

executed on MentDB

Match a typed product name to the catalog

A user types a product name with small differences from the catalog. The script measures the edit distance and the sentence distance, picks the best catalog label, then cuts the SKU into family and variant and checks its parts.

Pipeline

01 · editEdit distancestring levenshtein_distance→02 · sentenceSentence distancestring sentence_distance→03 · bestBest label in the catalogstring sentences_distance→04 · cutCut the SKUstring substr→05 · checkCharacter classesstring is_alpha_num

Script

admin

#1. A product name typed by a user, to match against the catalog;
-> "[typed]" "Ecran 27 pouces IPS";
-> "[ref]" "SKU-EC27-IPS";
#2. Edit distance and sentence distance to the catalog labels;
-> "[lev]" (string levenshtein_distance (string lower "Ecran 27p IPS") (string lower "ecran 27 ips"));
-> "[sd]" (string sentence_distance 40 70 [typed] "Ecran IPS 27 pouces noir");
-> "[best]" (string sentences_distance 40 70 [typed] "{\"1\": {\"i\": \"Clavier sans fil\", \"o\": \"SKU-KB01\"}, \"2\": {\"i\": \"Ecran IPS 27 pouces\", \"o\": \"SKU-EC27-IPS\"}, \"3\": {\"i\": \"Ecran 24 pouces TN\", \"o\": \"SKU-EC24-TN\"}}");
-> "[sku]" (json load select "rx_best" [best] "/best_predict");
json unload "rx_best";
#3. Cut the SKU into its parts;
-> "[p1]" (string instr [ref] "-");
-> "[p2]" (string locate [ref] "-" (+ [p1] 1));
concat "{\"levenshtein\": " [lev] ", \"sentence_distance\": " [sd] ", \"best_sku\": \"" [sku] "\", \"family\": \"" (string substr [ref] (+ [p1] 1) [p2]) "\", \"variant\": \"" (string substring [ref] (+ [p2] 1)) "\", \"prefix\": \"" (string left [ref] 3) "\", \"suffix\": \"" (string right [ref] 3) "\", \"dash_pos\": [" (string position [ref] "-") ", " (string strpos [ref] "-" 4) "], \"sorted\": " (string order "[\"SKU-EC27-IPS\", \"SKU-EC24-TN\", \"SKU-KB01\"]") ", \"cmp\": " (string strcmp "SKU-EC24-TN" "SKU-EC27-IPS") ", \"reversed\": \"" (string reverse [ref]) "\", \"grouped\": \"" (string repeat_insert_str "4111111111111111" " " 4) "\", \"letters_only\": " (string is_letter "Ecran") ", \"alnum\": " (string is_alpha_num "EC27") ", \"alnum_uds\": " (string is_alpha_num_uds "EC_27") ", \"digits\": " (string is_number_char "27") "}"

Result

mentdb

{"levenshtein": 1, "sentence_distance": 200.0, "best_sku": "SKU-EC27-IPS", "family": "EC27", "variant": "IPS", "prefix": "SKU", "suffix": "IPS", "dash_pos": [3, 8], "sorted": ["SKU-EC24-TN","SKU-EC27-IPS","SKU-KB01"], "cmp": -3, "reversed": "SPI-72CE-UKS", "grouped": "4111 1111 1111 1111 ", "letters_only": 1, "alnum": 1, "alnum_uds": 1, "digits": 1}

Go further

Raise the activation percent of string sentences_distance to be stricter
Sort candidate SKUs with string order before showing them

executed on MentDB

Mail merge: fill a template from JSON data

Fill the {{placeholders}} of a shipping mail with the data of an order: values are first formatted the French way (capital first letter, 1 249,50 with a thousands space, dd/MM/yyyy), then merged with a loop on the JSON fields. Unresolved placeholders are counted and removed, and the text is produced in plain and in safe HTML.

Pipeline

01 · formatThousands separatorstring repeat_insert_str→02 · dateFrench datedate format→03 · mergeOne replace per fieldjson parse_obj→04 · leftPlaceholders without datastring regexp_count→05 · htmlHTML versionstring html_escape

Script

admin

#1. A mail template with placeholders and the data of one customer;
-> "[tpl]" "Hello {{first_name}},|your order {{order}} of {{amount}} EUR ships on {{ship_date}}.|Track it: {{track_url}}|{{missing}}";
json load "rx_p4l_data" "{\"first_name\": \"élodie\", \"order\": \"A-1042\", \"amount\": 1249.5, \"ship_date\": \"2026-11-03\", \"track_url\": \"https://track.example.com/?id=A-1042&lang=fr\"}";
#2. Format the values the French way before the merge: 1 249,50 and dd/MM/yyyy;
json uobject "rx_p4l_data" "/" "first_name" (string first_letter_upper (json select "rx_p4l_data" "/first_name")) STR;
-> "[amt]" (json select "rx_p4l_data" "/amount");
-> "[int]" (string trim (string reverse (string repeat_insert_str (string reverse (math trunc [amt])) " " 3)));
-> "[cents]" (string lpad (math round (* (math frac_part [amt]) 100) 0) "0" 2);
json uobject "rx_p4l_data" "/" "amount" (concat [int] "," [cents]) STR;
json uobject "rx_p4l_data" "/" "ship_date" (date format (json select "rx_p4l_data" "/ship_date") "yyyy-MM-dd" "dd/MM/yyyy") STR;
#3. Replace each placeholder by its value;
-> "[text]" [tpl];
json parse_obj "rx_p4l_data" "/" "[k]" "[v]" {
	-> "[text]" (string replace [text] (concat "{{" [k] "}}") [v]);
};
json unload "rx_p4l_data";
#4. Placeholders without data are listed then removed;
-> "[left]" (string regexp_count [text] "\{\{[a-z_]+\}\}");
-> "[name]" (string regexp_extract [text] "\{\{([a-z_]+)\}\}" 1);
-> "[text]" (string regexp_replace [text] "\|?\{\{[a-z_]+\}\}" "");
#5. Plain text for the SMS, HTML for the mail;
-> "[plain]" (string replace [text] "|" [_n_]);
-> "[html]" (concat "<p>" (string replace (string html_escape [text]) "|" "<br>") "</p>");
concat [plain] [_n_] "--- " (string length [plain]) " chars, unresolved: " [left] " (" [name] ")" [_n_] [html];

Result

mentdb

Hello Élodie,
your order A-1042 of 1 249,50 EUR ships on 03/11/2026.
Track it: https://track.example.com/?id=A-1042&lang=fr
--- 123 chars, unresolved: 1 (missing)
<p>Hello Élodie,<br>your order A-1042 of 1 249,50 EUR ships on 03/11/2026.<br>Track it: https://track.example.com/?id=A-1042&amp;lang=fr</p>

Go further

Keep the templates in a node or in translate set keys to have one version per language.
Send the HTML with mail send once the text has no {{ left.

executed on MentDB

Mask personal data before writing a log line

A support log line contains an e-mail, a card number, an IBAN, a phone, an IP and an API token. The script masks each one with the rule of its kind (PCI DSS keeps the last 4 digits of a card), redacts the secret and replaces the user by a stable salted pseudonym.

Pipeline

01 · findValues in the linestring regexp_extract→02 · mailFirst letter and domainstring repeat→03 · cardLast 4 digits, groupedstring repeat_insert_str→04 · maskIBAN, phone, IP, tokenstring regexp_replace→05 · pseudoStable pseudonymstring sha_256

Script

admin

#1. A support log line full of personal data, before it goes to the shared logs;
-> "[line]" "2026-11-01 09:12 user=marie.dupont@gmail.com card=4970 1012 3456 7890 iban=FR7630006000011234567890189 tel=+33 6 12 34 56 78 ip=192.168.1.42 token=sk_live_9f8e7d6c5b4a";
#2. E-mail: keep the first letter and the domain;
-> "[mail]" (string regexp_extract [line] "user=([^ ]+)" 1);
-> "[mail_m]" (concat (string left [mail] 1) (string repeat "*" (- (string instr [mail] "@") 1)) (string substring [mail] (string instr [mail] "@")));
#3. Card: only the last 4 digits, the PCI DSS rule;
-> "[card]" (string regexp_replace (string regexp_extract [line] "card=([0-9 ]{19})" 1) " " "");
-> "[card_m]" (string repeat_insert_str (concat (string repeat "*" 12) (string right [card] 4)) " " 4);
#4. IBAN, phone, IP and secret with regular expressions;
-> "[out]" (string replace [line] [mail] [mail_m]);
-> "[out]" (string regexp_replace [out] "card=[0-9 ]{19}" (concat "card=" (string trim [card_m])));
-> "[out]" (string regexp_replace [out] "(FR[0-9]{2})[0-9]{19}([0-9]{4})" "$1*******************$2");
-> "[out]" (string regexp_replace [out] "(\+33 [0-9]) [0-9]{2} [0-9]{2} [0-9]{2} ([0-9]{2})" "$1 ** ** ** $2");
-> "[out]" (string regexp_replace [out] "ip=([0-9]+\.[0-9]+)\.[0-9]+\.[0-9]+" "ip=$1.x.x");
-> "[out]" (string regexp_replace [out] "token=[A-Za-z0-9_]+" "token=[REDACTED]");
#5. A stable pseudonym lets support follow the same user without the e-mail;
-> "[pseudo]" (concat "usr_" (string left (string sha_256 (concat "salt-2026:" (string lower [mail]))) 10));
concat [out] [_n_] "pseudonym: " [pseudo] [_n_] "digits left in clear: " (string regexp_count [out] "[0-9]") " of " (string regexp_count [line] "[0-9]") ", e-mail kept: " (string contains [out] [mail]) ", same length: " (== (string length (string trim [card_m])) 19);

Result

mentdb

2026-11-01 09:12 user=m***********@gmail.com card=**** **** **** 7890 iban=FR76*******************0189 tel=+33 6 ** ** ** 78 ip=192.168.x.x token=[REDACTED]
pseudonym: usr_a8895a160d
digits left in clear: 33 of 79, e-mail kept: 0, same length: 1

Go further

Put the masking in an exe script and include it in every log write of your services.
Store the salt in a private parameter (parameter set_private) rather than in the code.

executed on MentDB

Write and read a fixed width bank transfer file

Build a fixed width transfer file for a bank: detail records with a sequence number, a name in upper case without accents cut or padded to 24 characters, an 11-digit account and the amount in cents on 12 digits, then a trailer with the count and the control total, and an MD5 checksum. The bank side reads the fields back by position and checks the total.

Pipeline

01 · nameUpper case, no accents, 24 charsstring unaccent→02 · padFixed widthsstring rpad→03 · zerosAmount in cents on 12 digitsstring lpad→04 · sumFile checksumstring md5→05 · readFields by positionstring substring

Script

admin

#1. Transfers to send to the bank in a fixed width file: type 2, name 24, account 11, amount 12 in cents;
-> "[rows]" "Hélène Côté-Dubois;00012345678;1250.5|Jean Martin;98765432100;89.99|Société Réunionnaise des Eaux Tropicales;55511122233;14000";
-> "[file]" "";
-> "[total]" 0;
for (-> "[i]" 1) (<= [i] (atom size [rows] "|")) (++ "[i]") {
	-> "[r]" (atom get [rows] [i] "|");
	#2. Banks want upper case without accents, cut or padded to the width;
	-> "[name]" (string rpad (string left (string upper (string unaccent (atom get [r] 1 ";"))) 24) " " 24);
	-> "[cents]" (math round (* (atom get [r] 3 ";") 100) 0);
	-> "[total]" (+ [total] [cents]);
	-> "[file]" (concat [file] "02" (string lpad [i] "0" 6) [name] (atom get [r] 2 ";") (string lpad [cents] "0" 12) [_n_]);
};
#3. Trailer record with the count and the control total, then a checksum;
-> "[file]" (concat [file] "09" (string lpad (atom size [rows] "|") "0" 6) (string space 35) (string lpad [total] "0" 12));
-> "[md5]" (string md5 [file]);
#4. The bank side: read the fields back by position and check the total;
-> "[check]" 0;
-> "[parsed]" "";
for (-> "[j]" 1) (< [j] (atom size [file] [_n_])) (++ "[j]") {
	-> "[l]" (atom get [file] [j] [_n_]);
	-> "[amount]" (/ (string substring [l] 43 55) 100);
	-> "[check]" (+ [check] (string substring [l] 43 55));
	-> "[parsed]" (concat [parsed] (string rtrim (string substring [l] 8 32)) " | " (string substring [l] 32 43) " | " [amount] [_n_]);
};
-> "[trailer]" (atom get_last [file] [_n_]);
concat [file] [_n_] "--- every line " (string length [trailer]) " chars, md5 " [md5] [_n_] [parsed] "control total: " (string lrtrim0 (string right [trailer] 12)) " cents, recomputed: " [check] ", equal: " (== [check] (string right [trailer] 12));

Result

mentdb

02000001HELENE COTE-DUBOIS      00012345678000000125050
02000002JEAN MARTIN             98765432100000000008999
02000003SOCIETE REUNIONNAISE DES55511122233000001400000
09000003                                   000001534049
--- every line 55 chars, md5 b9294a861c1315974513bbdc5990d59e
HELENE COTE-DUBOIS | 00012345678 | 1250.5
JEAN MARTIN | 98765432100 | 89.99
SOCIETE REUNIONNAISE DES | 55511122233 | 14000
control total: 1534049 cents, recomputed: 1534049, equal: 1

Go further

Write the file with file create "tmp/..." and send it with sftp put in a job.
Real bank formats (CFONB 160, AFB 120) only change the widths and the record types.