phrase search in meteor search-source package - regex

I have a meteor app for which I added the search-source package to search certain collections and it works partially. That is, when I search for the term foo bar it returns results for each of "foo" and "bar". This is fine, but I want to also be able to wrap the terms in quotes this way: "foo bar" and get results for an exact match only. at the moment when i do this i get an empty set. Here is my server code:
//Server.js
SearchSource.defineSource('FruitBasket', function(searchText, options) {
// options = options || {}; // to be sure that options is at least an empty object
if(searchText) {
var regExp = buildRegExp(searchText);
var selector = {$or: [
{'fruit.name': regExp},
{'fruit.season': regExp},
{'fruit.treeType': regExp}
]};
return Basket.find(selector, options).fetch();
} else {
return Basket.find({}, options).fetch();
}
});
function buildRegExp(searchText) {
// this is a dumb implementation
var parts = searchText.trim().split(/[ \-\:]+/);
return new RegExp("(" + parts.join('|') + ")", "ig");
}
and my client code:
//Client.js
Template.dispResults.helpers({
getPackages_fruit: function() {
return PackageSearch_fruit.getData({
transform: function(matchText, regExp) {
return matchText.replace(regExp, "<b>$&</b>")
},
sort: {isoScore: -1}
});
}
});
Thanks in advance!

I've modified the .split pattern so that it ignores everything between double quotes.
/[ \-\:]+(?=(?:[^\"]*\"[^\"]*\")*[^\"]*$)/
Thus, you can simply wrap an exact phrase search in double quotes and it won't get split.
There is one more thing; since we don't need the quotes, they are removed in the next line using a .map function with a regex that replaces double quotes at the start or the end of a string part: /^"|"$/
Sample code:
function buildRegExp(searchText) {
// exact phrase search in double quotes won't get split
var arr = searchText.trim().split(/[ \-\:]+(?=(?:[^\"]*\"[^\"]*\")*[^\"]*$)/);
var parts = arr.map(function(x){return x.replace(/^"|"$/g, '');});
return new RegExp("(" + parts.join('|') + ")", "ig");
}
console.log(buildRegExp("foo bar"));
console.log(buildRegExp("\"foo bar\""));

Related

Typescript regex exclude whole string if followed by specific string

I'm been running into weird issues with regex and Typescript in which I'm trying to have my expression replace the value of test minus the first instance if followed by test. In other words, replace the first two lines that have test but for the third line below, replace only the second value of test.
[test]
[test].[db]
[test].[test]
Where it should look like:
[newvalue]
[newvalue].[db]
[test].[newvalue]
I've come up with lots of variations but this is the one that I thought was simple enough to solve it and regex101 can confirm this works:
\[(\w+)\](?!\.\[test\])
But when using Typescript (custom task in VSTS build), it actually replaces the values like this:
[newvalue]
[newvalue].[db]
[newvalue].[test]
Update: It looks like a regex like (test)(?!.test) breaks when changing the use cases removing the square brackets, which makes me think this might be somewhere in the code. Could the problem be with the index that the value is replaced at?
Some of the code in Typescript that is calling this:
var filePattern = tl.getInput("filePattern", true);
var tokenRegex = tl.getInput("tokenRegex", true);
for (var i = 0; i < files.length; i++) {
var file = files[i];
console.info(`Starting regex replacement in [${file}]`);
var contents = fs.readFileSync(file).toString();
var reg = new RegExp(tokenRegex, "g");
// loop through each match
var match: RegExpExecArray;
// keep a separate var for the contents so that the regex index doesn't get messed up
// by replacing items underneath it
var newContents = contents;
while((match = reg.exec(contents)) !== null) {
var vName = match[1];
// find the variable value in the environment
var vValue = tl.getVariable(vName);
if (typeof vValue === 'undefined') {
tl.warning(`Token [${vName}] does not have an environment value`);
} else {
newContents = newContents.replace(match[0], vValue);
console.info(`Replaced token [${vName }]`);
}
}
}
Full code is for the task I'm using this with: https://github.com/colindembovsky/cols-agent-tasks/blob/master/Tasks/ReplaceTokens/replaceTokens.ts
For me this regex is working like you are expecting:
\[(test)\](?!\.\[test\])
with a Typescript code like that
myString.replace(/\[(test)\](?!\.\[test\])/g, "[newvalue]");
Instead, the regex you are using should replace also the [db] part.
I've tried with this code:
class Greeter {
myString1: string;
myString2: string;
myString3: string;
greeting: string;
constructor(str1: string, str2: string, str3: string) {
this.myString1 = str1.replace(/\[(test)\](?!\.\[test\])/g, "[newvalue]");
this.myString2 = str2.replace(/\[(test)\](?!\.\[test\])/g, "[newvalue]");
this.myString3 = str3.replace(/\[(test)\](?!\.\[test\])/g, "[newvalue]");
this.greeting = this.myString1 + "\n" + this.myString2 + "\n" + this.myString3;
}
greet() {
return "Hello, these are your replacements:\n" + this.greeting;
}
}
let greeter = new Greeter("[test]", "[test].[db]", "[test].[test]");
let button = document.createElement('button');
button.textContent = "Say Hello";
button.onclick = function() {
alert(greeter.greet());
}
document.body.appendChild(button);
Online playground here.

Evaluate es6 template literals without eval() and new Function [duplicate]

Is it possible to create a template string as a usual string,
let a = "b:${b}";
and then convert it into a template string,
let b = 10;
console.log(a.template()); // b:10
without eval, new Function and other means of dynamic code generation?
In my project I've created something like this with ES6:
String.prototype.interpolate = function(params) {
const names = Object.keys(params);
const vals = Object.values(params);
return new Function(...names, `return \`${this}\`;`)(...vals);
}
const template = 'Example text: ${text}';
const result = template.interpolate({
text: 'Foo Boo'
});
console.log(result);
As your template string must get reference to the b variable dynamically (in runtime), so the answer is: NO, it's impossible to do it without dynamic code generation.
But, with eval it's pretty simple:
let tpl = eval('`'+a+'`');
No, there is not a way to do this without dynamic code generation.
However, I have created a function which will turn a regular string into a function which can be provided with a map of values, using template strings internally.
Generate Template String Gist
/**
* Produces a function which uses template strings to do simple interpolation from objects.
*
* Usage:
* var makeMeKing = generateTemplateString('${name} is now the king of ${country}!');
*
* console.log(makeMeKing({ name: 'Bryan', country: 'Scotland'}));
* // Logs 'Bryan is now the king of Scotland!'
*/
var generateTemplateString = (function(){
var cache = {};
function generateTemplate(template){
var fn = cache[template];
if (!fn){
// Replace ${expressions} (etc) with ${map.expressions}.
var sanitized = template
.replace(/\$\{([\s]*[^;\s\{]+[\s]*)\}/g, function(_, match){
return `\$\{map.${match.trim()}\}`;
})
// Afterwards, replace anything that's not ${map.expressions}' (etc) with a blank string.
.replace(/(\$\{(?!map\.)[^}]+\})/g, '');
fn = Function('map', `return \`${sanitized}\``);
}
return fn;
}
return generateTemplate;
})();
Usage:
var kingMaker = generateTemplateString('${name} is king!');
console.log(kingMaker({name: 'Bryan'}));
// Logs 'Bryan is king!' to the console.
Hope this helps somebody. If you find a problem with the code, please be so kind as to update the Gist.
What you're asking for here:
//non working code quoted from the question
let b=10;
console.log(a.template());//b:10
is exactly equivalent (in terms of power and, er, safety) to eval: the ability to take a string containing code and execute that code; and also the ability for the executed code to see local variables in the caller's environment.
There is no way in JS for a function to see local variables in its caller, unless that function is eval(). Even Function() can't do it.
When you hear there's something called "template strings" coming to JavaScript, it's natural to assume it's a built-in template library, like Mustache. It isn't. It's mainly just string interpolation and multiline strings for JS. I think this is going to be a common misconception for a while, though. :(
There are many good solutions posted here, but none yet which utilizes the ES6 String.raw method. Here is my contriubution. It has an important limitation in that it will only accept properties from a passed in object, meaning no code execution in the template will work.
function parseStringTemplate(str, obj) {
let parts = str.split(/\$\{(?!\d)[\wæøåÆØÅ]*\}/);
let args = str.match(/[^{\}]+(?=})/g) || [];
let parameters = args.map(argument => obj[argument] || (obj[argument] === undefined ? "" : obj[argument]));
return String.raw({ raw: parts }, ...parameters);
}
let template = "Hello, ${name}! Are you ${age} years old?";
let values = { name: "John Doe", age: 18 };
parseStringTemplate(template, values);
// output: Hello, John Doe! Are you 18 years old?
Split string into non-argument textual parts. See regex.
parts: ["Hello, ", "! Are you ", " years old?"]
Split string into property names. Empty array if match fails.
args: ["name", "age"]
Map parameters from obj by property name. Solution is limited by shallow one level mapping. Undefined values are substituted with an empty string, but other falsy values are accepted.
parameters: ["John Doe", 18]
Utilize String.raw(...) and return result.
TLDR:
https://jsfiddle.net/bj89zntu/1/
Everyone seems to be worried about accessing variables. Why not just pass them? I'm sure it won't be too hard to get the variable context in the caller and pass it down. Use
ninjagecko's answer to get the props from obj.
function renderString(str,obj){
return str.replace(/\$\{(.+?)\}/g,(match,p1)=>{return index(obj,p1)})
}
Here is the full code:
function index(obj,is,value) {
if (typeof is == 'string')
is=is.split('.');
if (is.length==1 && value!==undefined)
return obj[is[0]] = value;
else if (is.length==0)
return obj;
else
return index(obj[is[0]],is.slice(1), value);
}
function renderString(str,obj){
return str.replace(/\$\{.+?\}/g,(match)=>{return index(obj,match)})
}
renderString('abc${a}asdas',{a:23,b:44}) //abc23asdas
renderString('abc${a.c}asdas',{a:{c:22,d:55},b:44}) //abc22asdas
The issue here is to have a function that has access to the variables of its caller. This is why we see direct eval being used for template processing. A possible solution would be to generate a function taking formal parameters named by a dictionary's properties, and calling it with the corresponding values in the same order. An alternative way would be to have something simple as this:
var name = "John Smith";
var message = "Hello, my name is ${name}";
console.log(new Function('return `' + message + '`;')());
And for anyone using Babel compiler we need to create closure which remembers the environment in which it was created:
console.log(new Function('name', 'return `' + message + '`;')(name));
I liked s.meijer's answer and wrote my own version based on his:
function parseTemplate(template, map, fallback) {
return template.replace(/\$\{[^}]+\}/g, (match) =>
match
.slice(2, -1)
.trim()
.split(".")
.reduce(
(searchObject, key) => searchObject[key] || fallback || match,
map
)
);
}
Similar to Daniel's answer (and s.meijer's gist) but more readable:
const regex = /\${[^{]+}/g;
export default function interpolate(template, variables, fallback) {
return template.replace(regex, (match) => {
const path = match.slice(2, -1).trim();
return getObjPath(path, variables, fallback);
});
}
//get the specified property or nested property of an object
function getObjPath(path, obj, fallback = '') {
return path.split('.').reduce((res, key) => res[key] || fallback, obj);
}
Note: This slightly improves s.meijer's original, since it won't match things like ${foo{bar} (the regex only allows non-curly brace characters inside ${ and }).
UPDATE: I was asked for an example using this, so here you go:
const replacements = {
name: 'Bob',
age: 37
}
interpolate('My name is ${name}, and I am ${age}.', replacements)
#Mateusz Moska, solution works great, but when i used it in React Native(build mode), it throws an error: Invalid character '`', though it works when i run it in debug mode.
So i wrote down my own solution using regex.
String.prototype.interpolate = function(params) {
let template = this
for (let key in params) {
template = template.replace(new RegExp('\\$\\{' + key + '\\}', 'g'), params[key])
}
return template
}
const template = 'Example text: ${text}',
result = template.interpolate({
text: 'Foo Boo'
})
console.log(result)
Demo: https://es6console.com/j31pqx1p/
NOTE: Since I don't know the root cause of an issue, i raised a ticket in react-native repo, https://github.com/facebook/react-native/issues/14107, so that once they can able to fix/guide me about the same :)
You can use the string prototype, for example
String.prototype.toTemplate=function(){
return eval('`'+this+'`');
}
//...
var a="b:${b}";
var b=10;
console.log(a.toTemplate());//b:10
But the answer of the original question is no way.
I required this method with support for Internet Explorer. It turned out the back ticks aren't supported by even IE11. Also; using eval or it's equivalent Function doesn't feel right.
For the one that notice; I also use backticks, but these ones are removed by compilers like babel. The methods suggested by other ones, depend on them on run-time. As said before; this is an issue in IE11 and lower.
So this is what I came up with:
function get(path, obj, fb = `$\{${path}}`) {
return path.split('.').reduce((res, key) => res[key] || fb, obj);
}
function parseTpl(template, map, fallback) {
return template.replace(/\$\{.+?}/g, (match) => {
const path = match.substr(2, match.length - 3).trim();
return get(path, map, fallback);
});
}
Example output:
const data = { person: { name: 'John', age: 18 } };
parseTpl('Hi ${person.name} (${person.age})', data);
// output: Hi John (18)
parseTpl('Hello ${person.name} from ${person.city}', data);
// output: Hello John from ${person.city}
parseTpl('Hello ${person.name} from ${person.city}', data, '-');
// output: Hello John from -
I currently can't comment on existing answers so I am unable to directly comment on Bryan Raynor's excellent response. Thus, this response is going to update his answer with a slight correction.
In short, his function fails to actually cache the created function, so it will always recreate, regardless of whether it's seen the template before. Here is the corrected code:
/**
* Produces a function which uses template strings to do simple interpolation from objects.
*
* Usage:
* var makeMeKing = generateTemplateString('${name} is now the king of ${country}!');
*
* console.log(makeMeKing({ name: 'Bryan', country: 'Scotland'}));
* // Logs 'Bryan is now the king of Scotland!'
*/
var generateTemplateString = (function(){
var cache = {};
function generateTemplate(template){
var fn = cache[template];
if (!fn){
// Replace ${expressions} (etc) with ${map.expressions}.
var sanitized = template
.replace(/\$\{([\s]*[^;\s\{]+[\s]*)\}/g, function(_, match){
return `\$\{map.${match.trim()}\}`;
})
// Afterwards, replace anything that's not ${map.expressions}' (etc) with a blank string.
.replace(/(\$\{(?!map\.)[^}]+\})/g, '');
fn = cache[template] = Function('map', `return \`${sanitized}\``);
}
return fn;
};
return generateTemplate;
})();
Still dynamic but seems more controlled than just using a naked eval:
const vm = require('vm')
const moment = require('moment')
let template = '### ${context.hours_worked[0].value} \n Hours worked \n #### ${Math.abs(context.hours_worked_avg_diff[0].value)}% ${fns.gt0(context.hours_worked_avg_diff[0].value, "more", "less")} than usual on ${fns.getDOW(new Date())}'
let context = {
hours_worked:[{value:10}],
hours_worked_avg_diff:[{value:10}],
}
function getDOW(now) {
return moment(now).locale('es').format('dddd')
}
function gt0(_in, tVal, fVal) {
return _in >0 ? tVal: fVal
}
function templateIt(context, template) {
const script = new vm.Script('`'+template+'`')
return script.runInNewContext({context, fns:{getDOW, gt0 }})
}
console.log(templateIt(context, template))
https://repl.it/IdVt/3
I made my own solution doing a type with a description as a function
export class Foo {
...
description?: Object;
...
}
let myFoo:Foo = {
...
description: (a,b) => `Welcome ${a}, glad to see you like the ${b} section`.
...
}
and so doing:
let myDescription = myFoo.description('Bar', 'bar');
I came up with this implementation and it works like a charm.
function interpolateTemplate(template: string, args: any): string {
return Object.entries(args).reduce(
(result, [arg, val]) => result.replace(`$\{${arg}}`, `${val}`),
template,
)
}
const template = 'This is an example: ${name}, ${age} ${email}'
console.log(interpolateTemplate(template,{name:'Med', age:'20', email:'example#abc.com'}))
You could raise an error if arg is not found in template
This solution works without ES6:
function render(template, opts) {
return new Function(
'return new Function (' + Object.keys(opts).reduce((args, arg) => args += '\'' + arg + '\',', '') + '\'return `' + template.replace(/(^|[^\\])'/g, '$1\\\'') + '`;\'' +
').apply(null, ' + JSON.stringify(Object.keys(opts).reduce((vals, key) => vals.push(opts[key]) && vals, [])) + ');'
)();
}
render("hello ${ name }", {name:'mo'}); // "hello mo"
Note: the Function constructor is always created in the global scope, which could potentially cause global variables to be overwritten by the template, e.g. render("hello ${ someGlobalVar = 'some new value' }", {name:'mo'});
You should try this tiny JS module, by Andrea Giammarchi, from github :
https://github.com/WebReflection/backtick-template
/*! (C) 2017 Andrea Giammarchi - MIT Style License */
function template(fn, $str, $object) {'use strict';
var
stringify = JSON.stringify,
hasTransformer = typeof fn === 'function',
str = hasTransformer ? $str : fn,
object = hasTransformer ? $object : $str,
i = 0, length = str.length,
strings = i < length ? [] : ['""'],
values = hasTransformer ? [] : strings,
open, close, counter
;
while (i < length) {
open = str.indexOf('${', i);
if (-1 < open) {
strings.push(stringify(str.slice(i, open)));
open += 2;
close = open;
counter = 1;
while (close < length) {
switch (str.charAt(close++)) {
case '}': counter -= 1; break;
case '{': counter += 1; break;
}
if (counter < 1) {
values.push('(' + str.slice(open, close - 1) + ')');
break;
}
}
i = close;
} else {
strings.push(stringify(str.slice(i)));
i = length;
}
}
if (hasTransformer) {
str = 'function' + (Math.random() * 1e5 | 0);
if (strings.length === values.length) strings.push('""');
strings = [
str,
'with(this)return ' + str + '([' + strings + ']' + (
values.length ? (',' + values.join(',')) : ''
) + ')'
];
} else {
strings = ['with(this)return ' + strings.join('+')];
}
return Function.apply(null, strings).apply(
object,
hasTransformer ? [fn] : []
);
}
template.asMethod = function (fn, object) {'use strict';
return typeof fn === 'function' ?
template(fn, this, object) :
template(this, fn);
};
Demo (all the following tests return true):
const info = 'template';
// just string
`some ${info}` === template('some ${info}', {info});
// passing through a transformer
transform `some ${info}` === template(transform, 'some ${info}', {info});
// using it as String method
String.prototype.template = template.asMethod;
`some ${info}` === 'some ${info}'.template({info});
transform `some ${info}` === 'some ${info}'.template(transform, {info});
Faz assim (This way):
let a = 'b:${this.b}'
let b = 10
function template(templateString, templateVars) {
return new Function('return `' + templateString + '`').call(templateVars)
}
result.textContent = template(a, {b})
<b id=result></b>
Since we're reinventing the wheel on something that would be a lovely feature in javascript.
I use eval(), which is not secure, but javascript is not secure. I readily admit that I'm not excellent with javascript, but I had a need, and I needed an answer so I made one.
I chose to stylize my variables with an # rather than an $, particularly because I want to use the multiline feature of literals without evaluating til it's ready. So variable syntax is #{OptionalObject.OptionalObjectN.VARIABLE_NAME}
I am no javascript expert, so I'd gladly take advice on improvement but...
var prsLiteral, prsRegex = /\#\{(.*?)(?!\#\{)\}/g
for(i = 0; i < myResultSet.length; i++) {
prsLiteral = rt.replace(prsRegex,function (match,varname) {
return eval(varname + "[" + i + "]");
// you could instead use return eval(varname) if you're not looping.
})
console.log(prsLiteral);
}
A very simple implementation follows
myResultSet = {totalrecords: 2,
Name: ["Bob", "Stephanie"],
Age: [37,22]};
rt = `My name is #{myResultSet.Name}, and I am #{myResultSet.Age}.`
var prsLiteral, prsRegex = /\#\{(.*?)(?!\#\{)\}/g
for(i = 0; i < myResultSet.totalrecords; i++) {
prsLiteral = rt.replace(prsRegex,function (match,varname) {
return eval(varname + "[" + i + "]");
// you could instead use return eval(varname) if you're not looping.
})
console.log(prsLiteral);
}
In my actual implementation, I choose to use #{{variable}}. One more set of braces. Absurdly unlikely to encounter that unexpectedly. The regex for that would look like /\#\{\{(.*?)(?!\#\{\{)\}\}/g
To make that easier to read
\#\{\{ # opening sequence, #{{ literally.
(.*?) # capturing the variable name
# ^ captures only until it reaches the closing sequence
(?! # negative lookahead, making sure the following
# ^ pattern is not found ahead of the current character
\#\{\{ # same as opening sequence, if you change that, change this
)
\}\} # closing sequence.
If you're not experienced with regex, a pretty safe rule is to escape every non-alphanumeric character, and don't ever needlessly escape letters as many escaped letters have special meaning to virtually all flavors of regex.
You can refer to this solution
const interpolate = (str) =>
new Function(`return \`${new String(str)}\`;`)();
const foo = 'My';
const obj = {
text: 'Hanibal Lector',
firstNum: 1,
secondNum: 2
}
const str = "${foo} name is : ${obj.text}. sum = ${obj.firstNum} + ${obj.secondNum} = ${obj.firstNum + obj.secondNum}";
console.log(interpolate(str));
I realize I am late to the game, but you could:
const a = (b) => `b:${b}`;
let b = 10;
console.log(a(b)); // b:10

regex search term in variable

Do you know how I am able to put the search string inside the RegExp?
Let say if my search term is 'Ame', then I can write
v.name.search(new RegExp(/Ame/i)). //It works
var search = $("#search-query").val();
v.name.search(new RegExp('/'+search+'/i' //It doesn't work
However if the value 'Ame' was stored in the var 'search', how do I use the var?
var search = $("#search-query").val();
if($("#search-query").val().length <1){
$scope.hiddenError = true;
return;
}
$.each(json.products, function(i, v) {
if (v.name.search(new RegExp('/'+search+'/i')) != -1) { //doesn't work
$scope.recentGame.push({ label:v.name, value: v.type, link: v.url });
return;
}
});
When you want to build a custom regular expression from a string, you don't include the delimiters / or the options. You use the following form of the function constructor:
var regex = new RegExp(search, "i");
and use that in the search method.

What is the equalient of JavaScript's "s.replace(/[^\w]+/g, '-')" in Dart language?

I am trying to get the following working code in JavaScript also working in Dart.
https://jsfiddle.net/8xyxy8jp/1/
var s = "We live, on the # planet earth";
var results = s.replace(/[^\w]+/g, '-');
document.getElementById("output").innerHTML = results;
Which gives the output
We-live-on-the-planet-earth
I have tried this Dart code
void main() {
print( "We live, on the # planet earth".replaceAll("[^\w]+","-"));
}
But the output becomes the same.
What am I missing here?
If you want replaceAll() to process the argument as regular expression you need to pass a RegExp instance. I usually use r as prefix for the regex string to make it a raw string where not interpolation ($, \, ...) takes place.
main() {
var s = "We live, on the # planet earth";
var result = s.replaceAll(new RegExp(r'[^\w]+'), '-');
print(result);
}
Try it in DartPad

Added plus sign before number input in angularjs

I am using this directive to keep user typing only number into input tag.
app.directive('validNumber', function () {
return {
require: '?ngModel',
link: function (scope, element, attrs, ngModelCtrl) {
if (!ngModelCtrl) {
return;
}
ngModelCtrl.$parsers.push(function (val) {
if (angular.isUndefined(val)) {
var val = '';
}
var clean = val.replace(/[^0-9\.]/g, '');
var decimalCheck = clean.split('.');
if (!angular.isUndefined(decimalCheck[0])) {
decimalCheck[0] = decimalCheck[0].slice(0, 10);
if (!angular.isUndefined(decimalCheck[1])) {
clean = decimalCheck[0] + '.' + decimalCheck[1];
}
else {
clean = decimalCheck[0];
}
//console.log(decimalCheck[0][0]);
}
if (!angular.isUndefined(decimalCheck[1])) {
decimalCheck[1] = decimalCheck[1].slice(0, 3);
clean = decimalCheck[0] + '.' + decimalCheck[1];
}
if (val !== clean) {
ngModelCtrl.$setViewValue(clean);
ngModelCtrl.$render();
}
return clean;
});
element.bind('keypress', function (event) {
if (event.keyCode === 32) {
event.preventDefault();
}
});
}
};
});
But now i want to custome this, that means user can type ONLY ONE of "+" or "-" in the first. I think i have to change this pattern of
var clean = val.replace(/[^0-9\.]/g, '');
i also try to change into val.replace(/[^0-9.+-]/g, ''). It works but incorrectly, with this pattern user can type more "+" and "-" in any position of input field. I just wanna keep user typing ONLY ONE of "+" or "-" in the first like "+1234" or "-1234"
This is more of a regex problem than an AngularJS one, so you might have more luck there: https://stackoverflow.com/questions/tagged/regex
I'll try help you though. I think the regex you want matches a single +-, then any number of digits, then optionally a decimal point, then any number of digits. A single regex to match that is:
^[+-]?[0-9]*\.?[0-9]*
Have a read about groups and the '?' operator. This regex allows:
+.
-.
which don't make sense as input. You could design clever regexes to omit those results, but I think it would be easier to check the entry programmatically.
Finally, there are also very likely regexes online to help you solve any regex problem you ever come across more comprehensivley than you could. Just google an english description next time, and check out this for what you want:
http://www.regular-expressions.info/floatingpoint.html