Phone Numbers & SMS Parts

Accepted Tanzanian number formats, E.164 normalization, GSM-7 vs UCS-2 encoding, and how parts map to credits.

#Accepted formats

Only Tanzania mobile numbers are accepted for domestic sends. The API accepts several formats and normalizes to E.164 before dispatch — every SDK does the same locally before any network call:

Format	Example	Accepted
Tanzania local (07xx / 06xx)	`0712345678`	Yes — converted to `+255712345678`
International with +	`+255712345678`	Yes
International without +	`255712345678`	Yes
Other countries	`+254712345678`	No (Tanzania-only MVP)

Valid Tanzania mobile prefixes: 071 072 073 074 075 076 077 078. All of these are equivalent:

json

{ "to": "0712345678" }
{ "to": "+255712345678" }
{ "to": "255712345678" }

Invalid numbers fail fast with 400 invalid_phone — nothing is charged. For international destinations see SMS Rates; international sends are charged per part using the destination country's rate card.

#Encoding detection

The API auto-detects encoding from the characters in your message. Any character outside the GSM-7 set — emoji, Arabic, Chinese, curly quotes — flips the whole message to UCS-2:

Encoding	Single SMS	Per part (multipart)
GSM-7 (standard Latin)	160 chars	153 chars
UCS-2 (Arabic, Emoji, Chinese)	70 chars	67 chars

Multipart messages use 153/67 chars per part (not 160/70) because 7 bytes are reserved for concatenation headers.

#Parts → credits

You are charged 1 credit per SMS part by default (CREDIT_PRICE_PER_SMS_PART). Examples:

Message	Encoding	Chars	Parts	Credits
`Hello, your order is ready.`	GSM-7	28	1	1
OTP template (59 chars)	GSM-7	59	1	1
200-char promotional message	GSM-7	200	2	2
`Habari 😊 Bei yako ni 5000 TZS`	UCS-2	31	1	1
`مرحبا`	UCS-2	5	1	1

#Count parts before sending

parts.ts

typescript

import { getSmsPartInfo, detectEncoding } from "sendafrica";

const info = getSmsPartInfo("Hello, your order is ready.");
// { encoding: "GSM-7", length: 28, parts: 1, creditsRequired: 1 }

const emoji = getSmsPartInfo("Habari 😊 Bei yako ni 5000 TZS");
// { encoding: "UCS-2", length: 31, parts: 1, creditsRequired: 1 }

detectEncoding("Hello");     // "GSM-7"
detectEncoding("Hello 😊");  // "UCS-2"

Watch hidden unicode

Smart quotes (“…”), em-dashes (—), and zero-width characters silently switch a message to UCS-2 and cut your per-part budget from 160 to 70 chars. Sanitize user-composed text when possible.

Rate Limits & Idempotency

Sender IDs