Skip to content

Latest commit

 

History

History
163 lines (113 loc) · 3.4 KB

01-parsing-patterns.md

File metadata and controls

163 lines (113 loc) · 3.4 KB
Raw

Parsing Patterns

A cursor provides general interfaces to interact with the document. There are four basic functions that a cursor can perform:

  • Move
  • Test
  • Read
  • Backtrack

Move

To move a cursor, we can use next() or setIndex().

next() moves the cursor a number of characters. By using next(), we ensure that the cursor will move forward only:

const cursor = Cursor.from('Hello, World!')

cursor.next(2)
t.is(cursor.index, 2)

cursor.next(-1)
t.is(cursor.index, 2)

cursor.next(2)
t.is(cursor.index, 4)

setIndex() moves the cursor by setting its index directly.

const cursor = Cursor.from('Hello, World!')

cursor.setIndex(5)
t.is(cursor.index, 5)

cursor.setIndex(3)
t.is(cursor.index, 3)

This can be very helpful when working with regular expression:

const cursor = Cursor.from('-----1996-----')

const regexResult = cursor.exec(/[0-9]/)
cursor.setIndex(regexResult.index)

t.true(cursor.startsWith('1996'))

Test

In cursornext, there are three methods that can be used to test a cursor:

  • startsWith()
  • oneOf()
  • exec()

startsWith() takes an input string and tests whether the document from the cursor position starts with that string. For example:

1 | Hello, World!
  |        ^
  |        cursor

t.true(cursor.startsWith('World'))

oneOf() is similar to startsWith(), except it tests the cursor against multiple strings and returns the first string that matches. Otherwise, it would returns null.

1 | ----abcd-----
  |     ^
  |     cursor

t.is(cursor.oneOf(['bcd', 'cad', 'ab', 'abcd']), 'ab')
t.is(cursor.oneOf(['bca', 'bcd']), null)

exec() executes a search against the given regular expression. The search is executed on the document from the cursor position. It returns a RegExpExecArray or null:

1 | ----1234----
  |   ^
  |   cursor

const regexResult = cursor.exec(/[0-9]+/)

t.is(regexResult[0], '1234')
t.is(regexResult.index, 4)

If sticky flag is enabled, the match will only success if it starts from the cursor position in the document.

1 | ----1234----
  |   ^
  |   cursor

t.is(cursor.exec(/[0-9]+/y), null)
cursor.next(2)

const regexResult = cursor.exec(/[0-9]+/y)

t.is(regexResult[0], '1234')
t.is(regexResult.index, 4)

Read

In cursornext, the most common way to read the document is by using marker. We can create a marker from a cursor by using clone(), then read the text between two cursors by using takeUntil():

const cursor = Cursor.from('Hello, World!')

const marker = cursor.clone()
cursor.next(5)

// Read the input from the marker cursor position
// to the current cursor position.
const output = marker.takeUntil(cursor)

t.is(output, 'Hello')

exec() and lookahead() can also be used to extract information from the document:

const cursor = Cursor.from('Hello, World!')

t.is(cursor.lookahead(5), 'Hello')
cursor.next(5)

t.is(cursor.exec(/[a-zA-Z]+/)[0], 'World')

Backtrack

Sometimes, when working with complex syntax constraints, you may need to backtrack to the last checkpoint. The following example illustrates how to backtrack using moveTo() method:

// Mark the cursor position by cloning the cursor.
const marker = cursor.clone()

// Do something with the cursor!
// ...
// ...

// If some constraint are not sastisfied, backtrack
// the cursor to the marked position.
if (!sastisfied) {
  cursor.moveTo(marker)
}