Source text and words

Before the compiler asks what a program means, it cuts the file into words. Burmese is written without spaces, and the compiler has no word list, so a few plain rules decide where one word ends and the next begins. This page states those rules, from the the file itself up to the literals.

Source files

A Zaka program is a text file in UTF-8, saved with the extension .zaka. The compiler first puts the text into one standard form:

• a leading U+FEFF mark is dropped;
• CR LF and a lone CR each count as one line break;
• the text is normalized to Unicode NFC, and the signs of a Burmese vanna are put in one fixed order. The asat ် comes before the dot below ့, as in နှင့် and သည့်.

The last step is the reason two spellings that look the same lex the same. A name or a reserved word written with its signs in another order still matches.

Sentences and the full stop

A program is a list of statements. Each statement is a sentence that ends with the Burmese full stop ။. A line break does not end a statement, so one statement can cover several lines and several statements can share a line.

"တစ်" ဟု ပြသပါ။ "နှစ်" ဟု ပြသပါ။
၁ +
၂ ကို
ပြသပါ။

တစ်
နှစ်
၃

A missing ။ is reported as ZK0201 ("expected ။"). The list separator is ၊, which is used between arguments, parameters and the cells of a list literal; a comma is not a Zaka character.

ကိန်းရှင် ကိန်းများ သည် ကိန်းပြည့် စာရင်း ဖြစ်၍ တန်ဖိုးသည် ၁, ၂ တို့ပါသော စာရင်း ဖြစ်ပါစေ။

error[ZK0101]: invalid character in ","
  = help: lists are separated with "၊", not ","

Words and spaces

The lexer works in three steps: it cuts the text into vannas (Burmese letter clusters), digits and single other characters; it joins the pieces that touch into words; and it classifies each word. These rules decide the cuts.

• White space separates words. The seven white space characters are TAB, LINE FEED, VERTICAL TAB, FORM FEED, CARRIAGE RETURN, SPACE and ZERO WIDTH SPACE (U+200B, which the macOS Burmese input source enters for Shift + Space). One of them or a run of them count the same, and none is a token.
• Some characters are words on their own. ၊, ။, ၏ and each ASCII punctuation character except _ are cut out even with no space beside them, such as ( ) + - * / % ; , . and the quote marks. That is the reason ၁+၂ and ပြသပါ။ need no spaces. A pair that forms one sign (<= >= == != << >> && ||) is joined when its two characters touch.
• A name runs to the next white space or lone character. The name can mix scripts (အမှတ်_၁, age) as long as nothing separates the pieces.
• A word made of reserved words alone is split. ထက်ကြီးသည် is ထက် ကြီး သည်, and တန်ဖိုးကို is တန်ဖိုး ကို. The split picks the fewest pieces, and a word that holds just particles (such as ကိုကို) is not split.
• A word with a vanna that is not reserved is a name. အသက်ကို is one name, not အသက် then ကို.
• The words of a reserved phrase such as စားသော စားလဒ် or အောက်ပါတို့ကို လုပ်ဆောင်ပါ can have white space between the words or none; both read the same.

အသက် ကို ပြသပါ။     အသက်   |  ကို   |  ပြသပါ  |  ။       a name, two reserved words, a stop
အသက်ကို ပြသပါ။     အသက်ကို  |  ပြသပါ  |  ။                one name: no reserved split
ထက်ကြီးသည်          ထက်  |  ကြီး  |  သည်                  reserved words alone: split
၁+၂                 ၁  |  +  |  ၂                         signs are cut out

The fourth rule gives the one habit a Zaka writer needs: put a space between a name and the particle after it. When the joined word is not a known name, the compiler guesses that a space is missing and reports it:

ကိန်းရှင် အသက် သည် ကိန်းပြည့် ဖြစ်၍ တန်ဖိုးသည် ၂၀ ဖြစ်ပါစေ။
အသက်ကို ပြသပါ။

error[ZK0206]: "ပြသပါ" needs "ကို|ဟု" after operand 1, found no particle
help: write "အသက် ကို" with a space: a name joined to a particle is read as one word

A polite ending (သည်, ပါ, ပါသည်) written attached to a comparison verb, to မှန်, မမှန် or မဟုတ် is part of that word; the lexer keeps it as a note on the token. With a space, သည် is the subject marker instead. Natural wording tells which ending reads well where.

Comments

A comment starts at //, at ///, or at the word မှတ်ချက် ("note"), and runs to the end of the line. A comment is not a token. There is no comment that spans lines. A marker inside a string or character literal is just text, and မှတ်ချက် starts a comment just when it is a word on its own.

// two slashes
/// three slashes
မှတ်ချက် the word for "note"
"http://zaka.example // not a comment" ဟု ပြသပါ။   // a comment after a statement
၁ ကို ပြသပါ။ မှတ်ချက် also after a statement

http://zaka.example // not a comment
၁

Names

A name starts with a letter of a script, an emoji, _, or a Burmese character that is not a digit. After the first character it can also hold digits (Burmese or Western). Upper and lower case differ: Age and age are two names. A name cannot be just a reserved word or a row of reserved words.

ကိန်းရှင် အသက် သည် ကိန်းပြည့် ဖြစ်၍ တန်ဖိုးသည် ၁ ဖြစ်ပါစေ။
ကိန်းရှင် Age သည် ကိန်းပြည့် ဖြစ်၍ တန်ဖိုးသည် ၂ ဖြစ်ပါစေ။
ကိန်းရှင် age သည် ကိန်းပြည့် ဖြစ်၍ တန်ဖိုးသည် ၃ ဖြစ်ပါစေ။
ကိန်းရှင် အမှတ်_၁ သည် ကိန်းပြည့် ဖြစ်၍ တန်ဖိုးသည် ၄ ဖြစ်ပါစေ။
ကိန်းရှင် 🍎 သည် ကိန်းပြည့် ဖြစ်၍ တန်ဖိုးသည် ၅ ဖြစ်ပါစေ။
အသက် + Age + age + အမှတ်_၁ + 🍎 ကို ပြသပါ။

၁၅

Raw names

A name written between backticks is a raw name. It can hold reserved words, spaces and signs. It cannot hold a backtick or a line break, and it cannot be blank.

ကိန်းရှင် `ကို ပြသပါ` သည် ကိန်းပြည့် ဖြစ်၍ တန်ဖိုးသည် ၇ ဖြစ်ပါစေ။
ကိန်းရှင် `the total` သည် ကိန်းပြည့် ဖြစ်၍ တန်ဖိုးသည် ၉ ဖြစ်ပါစေ။
`ကို ပြသပါ` + `the total` ကို ပြသပါ။

၁၆

A raw name with no closing backtick on its line is ZK0108, and two backticks in a row are ZK0109. Writing a plain reserved word as a name gives ZK0201 ("expected a name").

ကိန်းရှင် `` သည် ကိန်းပြည့် ဖြစ်ပါစေ။

Numbers

A number literal is a run of digits. The digits are Burmese (၀ to ၉) or Western (0 to 9), but one number holds one script alone (ZK0103). A literal has up to three parts in a fixed order:

Part	Written	Notes
base	အခြေ ၁၆ ဖြင့်	optional; a base from 2 to 36, itself written in base 10 (ZK0218)
kind	လက္ခဏာမဲ့ ၈ ဘစ် ကိန်းပြည့်	optional; see Kinds and values (ZK0216, ZK0217)
digits	၂၅၅, FF, ၃.၁၄, 1.5e3	in base 10, a fraction . and an exponent are allowed

• Digits above 9 are the letters A to Z, in either case. A digit not below the base is ZK0219.
• A fraction needs digits on both sides of the point: ၁.၅, not ၁. or .၅.
• An exponent (e or E, an optional sign, digits) needs Western digits: 1.5e3 works, ၂.၅e3 is ZK0102. A fraction or an exponent makes a decimal number, and a base prefix cannot go with one.
• There is no negative literal. -၅ is the minus sign applied to ၅, and so is အနုတ် ၅. The compiler works it out before the program runs.
• A literal with no kind prefix is flexible: it takes the kind that its place wants (see Expressions). A literal that does not fit its kind is an error: ZK0220 when the kind is written in front of it, and ZK0406 when the place gave the kind.

၁၂၃ ကို ပြသပါ။                                    // Burmese digits
123 ကို ပြသပါ။                                     // Western digits: the same number
အခြေ ၂ ဖြင့် ၁၀၁၀ ကို ပြသပါ။                        // 1010 in base 2
အခြေ ၁၆ ဖြင့် ff ကို ပြသပါ။                          // letters, either case
အခြေ ၁၆ ဖြင့် လက္ခဏာမဲ့ ၈ ဘစ် ကိန်းပြည့် FF ကို ပြသပါ။   // base, then kind, then digits
၃.၁၄ ကို ပြသပါ။
1.5e3 ကို ပြသပါ။
အနုတ် ၅ ကို ပြသပါ။

၁၂၃
၁၂၃
၁၀
၂၅၅
၂၅၅
၃.၁၄
၁၅၀၀.၀
-၅

Numbers print with Burmese digits, whatever digits the program used.

၁၂.34 ကို ပြသပါ။

အခြေ ၂ ဖြင့် ၁၂ ကို ပြသပါ။

error[ZK0219]: "၂" is not a digit in base 2

Characters and strings

A character literal holds one Unicode scalar value in single quotes: 'က', 'A', '🔥'. A Burmese vanna with signs, such as ကို, is several scalar values, so 'ကို' is ZK0107. A string literal is text in double quotes. It has no kind of its own: it can be printed and nothing else (ZK0223). A literal must close on the line where it opens (ZK0104 for a string, ZK0105 for a character).

Escapes work in both. The backslash forms are for readers who know C, and the brace forms are the Burmese spelling. A brace is reserved inside a literal, so a plain { or } is doubled.

Escape	Means
{နောက်တစ်ကြောင်း} or \n	line feed
\r, \t, \0	carriage return, tab, the null character
\\, \', \"	a backslash, a single quote, a double quote
{{, }}	a brace {, a brace }
\uHHHH or \uHHHHHHHH	the code point with 4 or 8 hex digits
{ယူနီကုဒ် HHHH} or {ယူနီကုဒ် HHHHHHHH}	the same, in Burmese

A code point must be a scalar value: a surrogate, a value above 10FFFF and a digit count other than 4 or 8 are errors. Another {...} is ZK0106; braces are kept free so that a later version can give them a use.

"တစ်{နောက်တစ်ကြောင်း}နှစ်" ဟု ပြသပါ။
"a\\b" ဟု ပြသပါ။
"{{ }}" ဟု ပြသပါ။
'က' ကို ပြသပါ။
'{ယူနီကုဒ် 1000}' ကို ပြသပါ။
"x\u00000041y" ဟု ပြသပါ။

တစ်
နှစ်
a\b
{ }
က
က
xAy

"{မသိ}" ဟု ပြသပါ။

'ကို' ကို ပြသပါ။

error[ZK0107]: a character literal must hold one character, found 2

"မပိတ်ထားသော စာ ဟု ပြသပါ။

Reserved words

The words that Zaka gives a meaning to cannot be used as names. Reserved words lists each of them. A reserved word that ends in the written မည် also takes the spoken မယ်: ပြန်ပေးမည် and ပြန်ပေးမယ် are the same word. To use a reserved word as a name, put it between backticks.

Each diagnostic of this page has a code from ZK0101 to ZK0109. Error messages lists them with the others.